feat(agent-core-v2): let plugins contribute system prompt instructions via the manifest systemPrompt field - #2314
Merged
Merged
Conversation
…s via the manifest systemPrompt field
🦋 Changeset detectedLatest commit: 11c475e The changes in this PR will be included in the next version bump. This PR includes changesets to release 1 package
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
commit: |
…gin contributions at session scope - restore no longer re-renders or re-persists prompts: a resumed agent keeps its replayed profile binding (prompt and tool set) as persisted - a new Session-level convergence point reloads plugin skills into the session skill catalog before fanning out to every live agent prompt, and every catalog-kind plugin mutation awaits the whole pipeline; MCP-only toggles carry a distinct change kind and skip it - live refreshes after a restart re-resolve the bound profile by name and rebind the full slice (prompt, disallowed tools, active tools) atomically, warning and keeping the persisted state when the profile is gone; renders reuse the first-render timestamp and unchanged prompts are not re-persisted, so convergence never churns the wire - cap plugin system-prompt contributions (32 KB per field/file, 64 KB aggregate per prompt build) with manifest diagnostics and warnings - bump the changeset to minor: this is a new user-facing capability
…ssing-profile warning - add sessionPluginContribution to the domain-layer registry so lint:domain stays green - emit system-prompt-refresh-profile-missing once per profile name, matching the service's other deduped warnings - document the convergence timeout escape hatch and the klient exclusion of enabledSystemPrompts
…ion read failures - emit plugin-sections-oversized once per skipped-plugin signature - let enabledSystemPrompts failures propagate to the refresh catch (keeps the current prompt and warns) instead of silently rendering and persisting a prompt without plugin instructions - cover the convergence timeout cut-off with a fake-timers test - clarify that the first-render timestamp anchors per process
…load timing - run at most one convergence per session and bound each change's wait by the timeout, so a fan-out emitter never interleaves deliveries after a timed-out convergence - fire onDidReload as soon as the reload commits again, keeping hook reloads independent of prompt convergence - sign the plugin budget warning with an unambiguous key
…mantics - the timeout retry promise only holds once stalled work clears - note the per-session serial delivery cost model on the plugin change contract and the dual-queue invariant on the service
…ompt template
- place ${plugin_sections} on the same template line as
${skills_section} so prompts without either block render exactly as
before this feature
- note on the change contract that waitUntil work must not call back
into plugin mutations, and spell out the per-session convergence
order in the user docs
…bind it
- applyBindingSnapshot left the fork with no pinned profile, which
routed in-process forks into the post-restart catalog rebind and
could reset an inherited tool set; forks now inherit the source
agent's pinned profile object
- pin the first-render timestamp reuse with a ${now}-embedding test
and document the anchored ${now} semantics
- tighten the plugin docs budget and resume-refresh wording
- an agent created while a plugin convergence is in flight now waits for it, and a restored agent refreshes once after it, so a plugin mutation never straddles an agent's bootstrap - warn on a non-string systemPrompt field and strip a UTF-8 BOM from systemPromptPath files before trimming - correct the consumption-surface wording (every CLI surface on the experimental flag, not just kimi -p), the per-session queueing note, and the single-plugin combined budget clause
A permanently wedged convergence kept convergeTail pending forever, and the unconditional settled() wait in bindBootstrap would have blocked every later agent creation in that session; the join now races the shared convergence timeout and continues (a restored agent still refreshes once, which never touches the tail), and the timeout constant moves to the contract for reuse
…stores - a convergence fan-out could land while an agent's wire log is still replaying, dispatching a replay-visible config record whose effect the rest of the replay then overwrites; refreshSystemPrompt now skips while the wire restore is in progress - convergence completion is tracked by a generation counter; bootstrap compares it (after a bounded join) and refreshes a restored agent exactly once when a round completed after its creation began, replacing the wasConverging flag that could miss both windows
…nnot stop the pipeline - the fan-out now races the convergence timeout, so convergeTail always settles: a permanently hung refresh delays its round (blocked entries drain oldest-first on later changes) instead of killing the session's convergence for good - warn when agent bootstrap stops waiting on a stalled convergence - diagnose a blank systemPromptPath and pin the plugin-root escape guard with traversal, absolute-path, and symlink tests
…ys, roll the prompt clock daily - the convergence's skill-reload segment now races the same timeout as the fan-out, so no segment of the pipeline can wedge a session for good; it continues with the previous catalog and retries next change - a cold rebind that resets the tool set replays session-added user tools onto the new base instead of dropping them for the rest of the process - the rendered timestamp re-anchors when the UTC date rolls over, so long-lived processes keep a fresh clock while steady-state renders stay byte-stable within a day - the plugin budget warning dedupes per plugin id, and the docs note that systemPromptPath content is frozen until the next reload
… drift-free gate - restore replays the persisted binding untouched, then bootstrap refreshes only when drift-free inputs changed while the session was cold: the catalog profile's tool set/denylist, or the plugin-sections baseline persisted alongside the prompt on the existing bind/update payloads; directory-listing and date drift wait for live triggers, so quiet resumes append no replay-visible records - the rendered timestamp is day-precision (UTC date at 00:00, re-anchored on rollover), keeping steady-state renders byte-stable across resumes and sessions on the same day - consolidate both timeout helpers onto a shared raceOutcome, and drop the generation counter the gate supersedes - align the plugin-sections precedence prose with the AGENTS.md disclaimer (no self-granted authority, system instructions win on conflict)
…ons baseline - the gate's plugin-sections read now races the convergence timeout, so agent creation never blocks behind an unrelated plugin mutation - refreshes serialize per agent through a tail, so overlapping triggers cannot write prompts out of order - when plugin sections change but a plugin-free custom prompt does not, the new baseline lands as a sections-only update instead of making every later resume re-render in vain - align the system prompt's Date and Time paragraph with the day-precision anchored timestamp
Live sessions pick up plugin changes, while the default TUI and `kimi -p` paths ignore these fields. Signed-off-by: 7Sageer <sag77r@hotmail.com>
Plugin mutations still converge live agent prompts, but the session skill catalog goes back to refreshing only on explicit plugin reload, as before: the prompt feature does not need skill convergence, and the pre-existing manual-reload semantics stay uniform across all plugin contributions. Removes the convergence-driven skill reload, the reloadSource de-privatization, and their tests; restores the PluginSkillSource onDidReload forwarding and its catalog tests.
…xplicit reload Drop the live convergence machinery (the plugin onDidChange barrier, the sessionPluginContribution fan-out, the restored-prompt drift gate, and the day-precision render clock) so plugin system-prompt sections take effect at the same point as every other plugin contribution: /plugins reload or a new session. The profile now refreshes when the session skill catalog re-pulls its plugin source on reload, reading both the skill list and the prompt sections fresh.
…ia the manifest systemPrompt field
Pidbid
added a commit
to Pidbid/kkm
that referenced
this pull request
Jul 30, 2026
* feat(agent-core): custom agent files and secondary model on the v1 engine (MoonshotAI#2232) * feat(agent-core): custom agent files and secondary model on the v1 engine Migrate the custom agentfile and secondary-model capabilities from agent-core-v2 to the v1 engine so they work in the TUI and plain kimi -p sessions: - discover Markdown agent files from user/project/extra/explicit directories with the v2 precedence rules, a merged session profile catalog replacing the hardcoded builtin profile lookups, SYSTEM.md main prompt override, and ${base_prompt} backed by the effective default - --agent/--agent-file now work in print mode on the default engine; CreateSessionOptions gains agentProfile/agentFiles - [secondary_model] config + KIMI_SECONDARY_MODEL/EFFORT bind newly spawned subagents to a cheaper model behind the secondary-model experiment flag, with primary/secondary model params on Agent and AgentSwarm and upfront session warnings - full disallowedTools deny semantics (exact names + mcp__ globs) evaluated by the tool manager and persisted in the agent wire * fix(cli): guard optional agentFiles in the prompt runner runPrompt is also driven programmatically (headless goal flow) with options that never pass through the CLI parser defaults, so agentFiles can be undefined; mirror the addDirs optional-chaining pattern. Also extend the SDK experimental-feature assertion with the secondary-model flag. * fix(agent-core): preserve custom agent bindings on v1 * fix(agent-core): narrow secondary model error hints * fix(agent-core): persist custom agent profile bindings * Delete .changeset/sdk-agent-profile-options.md Signed-off-by: 7Sageer <sag77r@hotmail.com> * Update v1-custom-agent-files.md Signed-off-by: 7Sageer <sag77r@hotmail.com> * Update v1-secondary-model.md Signed-off-by: 7Sageer <sag77r@hotmail.com> * Update v1-custom-agent-files.md Signed-off-by: 7Sageer <sag77r@hotmail.com> * fix(agent-core): keep SYSTEM.md a prompt-only overlay for delegation * docs: update agent file and secondary model availability wording * fix(cli): reject --agent-file combined with session resume The resume path only forwards the agent file's name for the bound-profile assertion; the file's content is never re-applied (the session keeps its creation-time catalog snapshot). Previously the combination was silently accepted, so an edited file (or a same-named one) appeared to apply but did not. Reject it at option validation and document the constraint. * refactor(agent-core): share prompt-section prose and note v2 twins in agentfile headers The Windows notes, additional-dirs and skills prose blocks existed twice: inline in the builtin default template (system.md) and as constants in the agent-file renderer (from-file.ts). Extract them to profile/prompt-sections.ts as the single source: system.md renders them through injected KIMI_* template variables and from-file.ts imports the same constants. Rendered prompts are byte-identical for all four builtin profiles across macOS/Windows and skills/dirs on/off; a new test pins system.md to the shared constants. Also mark each profile/agentfile file with the path of its agent-core-v2 counterpart so format/semantics changes land in both engines. * feat(cli): add /secondary_model command for the subagent model Mirror /model: a picker with a thinking-effort step that persists [secondary_model] and live-applies to the current session via a new Session.setSecondaryModel RPC (node-sdk wrapper included), so newly spawned subagents bind the new model right away. The /model picker now hides the synthesized __secondary__ derived entry; docs and the update-config builtin skill mention the section. * feat(tui): show the bound model in subagent run stats Subagents report their model alias via agent.status.updated after spawn; resolve it to a display name and surface it in tool-call subagent stats and agent-group rows. * fix(agent-core): validate agent profile before session persistence * fix(agent-core): refresh subagent tools after model switch * fix(agent-core): show subagent model preferences * fix(agent-core): preserve secondary model recipe on live apply * fix(agent-core): make secondary model apply explicit * fix(tui): refresh secondary model display state * chore: merge secondary model changesets into one * Add /secondary_model command for subagent configuration Show each subagent's model in the subagent card header and agent-group rows. Requires the secondary-model experiment (KIMI_CODE_EXPERIMENTAL_SECONDARY_MODEL=1); run /secondary_model to pick a model and thinking effort, applied to the current session immediately. Signed-off-by: 7Sageer <sag77r@hotmail.com> * fix(agent-core): align explicit agent file precedence * fix(agent-core): let disallowedTools deny select_tools * chore(cli): drop engine mention from --agent/--agent-file help text * feat(cli): support --agent/--agent-file in the interactive TUI Bind the selected agent profile to the startup session when launching the TUI with --agent/--agent-file, including the session created after an OAuth login at startup. Sessions created later in the process (/new) keep the default profile. Make both flags creation-only in every mode: combining them with --session/--continue is now rejected in print mode too, since resume restores the bound agent from the session automatically. * fix(agent-core): persist new secondary-model selections under env overrides stripSecondaryModelConfig restored secondary_model.model/default_effort from raw whenever KIMI_SECONDARY_MODEL/KIMI_SECONDARY_EFFORT was set, so a /secondary_model pick made under the env vars was silently discarded on write. Restore from raw only when the value being written still equals the env value (an overlay round-trip), mirroring the pointer check in stripEnvModelConfig; a genuinely different selection now reaches config.toml. * fix(cli): report the effective secondary model when env overrides the pick /secondary_model toasted the picked alias even when KIMI_SECONDARY_MODEL/KIMI_SECONDARY_EFFORT made the session bind a different model. Read the effective binding back from the reloaded config (as /model does from session status) and warn with the env-overridden values instead. * feat(tui): show the bound model name in the AgentSwarm panel header --------- Signed-off-by: 7Sageer <sag77r@hotmail.com> (cherry picked from commit efac96c) * fix(node-sdk): wire applyPersistedSecondaryModel to agent-core-v2 (MoonshotAI#2345) * test(node-sdk): drop v1-only subagentNames from the resume parity projection Custom agent files made v1's resumed agent config carry the bound profile's delegatable subagent roster; v2's resumed agent state has no equivalent field, so the resume parity cases fail on main. Project the engine-owned field away instead of pinning it as a resume-data gap. * fix(node-sdk): wire applyPersistedSecondaryModel to agent-core-v2 On the v2 engine route the /secondary_model command persisted the recipe but failed to apply it to the current session: the SDK method fell through to the base class's not_implemented getRpc(). v1 pushes a reloaded config snapshot into the session because its spawn binding, tool descriptions, and cached startup warning all read that snapshot. agent-core-v2 resolves the secondary model live against IConfigService at spawn time and rebuilds the tool description per read, so the setConfig write already takes effect session-wide. The override keeps the rest of v1's contract: config reload, the same loud validations (session lookup, persist-first recipe check, pointed-model resolution wrapped at [secondary_model].model), and a warning-cache refresh via a new recheckSecondaryModelWarning on the session warning service. getSessionWarnings also surfaces the v2 secondary-model warning next to the AGENTS.md one, matching v1's aggregate. * fix(agent-core-v2): surface the subagent's bound model on status events The v2 model slice rides only the bind-time agent.status.updated, which precedes subagent.spawned and is dropped by clients that key child events off the spawn, so subagent cards never learned the model — and a single-step run emits no usage/context slice until it ends, so the model only appeared at completion. Re-affirm the binding right after the spawn announcement via a new IAgentProfileService.republishStatus, and fold a consistent usage/context/model snapshot into every status event at both v1 edges (kap-server's broadcaster and the in-process SDK session wiring, resolving the secondary-model derived id to a readable display name. EOF ) * Delete .changeset/subagent-card-model.md Signed-off-by: 7Sageer <sag77r@hotmail.com> * style(agent-core-v2): remove inline implementation comments --------- Signed-off-by: 7Sageer <sag77r@hotmail.com> (cherry picked from commit b850c5f) * docs: fix dead anchor links in en/zh docs (MoonshotAI#2348) * docs: fix dead anchor links in en/zh docs - #loop_control -> #loop-control (heading slug uses hyphens) - #secondary_model -> #secondary-model - env-vars model section anchors: kimi_model -> kimi-model - provider credential section anchors: configtoml -> config-toml - /provider management anchors: point at the renamed heading in each locale - hooks: point the stale config-files#hooks reference at the local Configuration section - en files: replace two leftover Chinese anchors with their English targets * docs: add missing .md extension to themes page links --------- Co-authored-by: qer <wbxl2000@outlook.com> (cherry picked from commit f8ec3d1) * feat(agent-core-v2): let plugins contribute system prompt instructions via the manifest systemPrompt field (MoonshotAI#2314) * feat(agent-core-v2): let plugins contribute system prompt instructions via the manifest systemPrompt field * feat(agent-core-v2): add systemPromptPath to load plugin system prompt from a file * docs: explain plugin system prompt templates * fix(agent-core-v2): refresh plugin system prompts after changes * fix(agent-core-v2): freeze restored profile bindings and converge plugin contributions at session scope - restore no longer re-renders or re-persists prompts: a resumed agent keeps its replayed profile binding (prompt and tool set) as persisted - a new Session-level convergence point reloads plugin skills into the session skill catalog before fanning out to every live agent prompt, and every catalog-kind plugin mutation awaits the whole pipeline; MCP-only toggles carry a distinct change kind and skip it - live refreshes after a restart re-resolve the bound profile by name and rebind the full slice (prompt, disallowed tools, active tools) atomically, warning and keeping the persisted state when the profile is gone; renders reuse the first-render timestamp and unchanged prompts are not re-persisted, so convergence never churns the wire - cap plugin system-prompt contributions (32 KB per field/file, 64 KB aggregate per prompt build) with manifest diagnostics and warnings - bump the changeset to minor: this is a new user-facing capability * fix(agent-core-v2): register the new session domain and dedupe the missing-profile warning - add sessionPluginContribution to the domain-layer registry so lint:domain stays green - emit system-prompt-refresh-profile-missing once per profile name, matching the service's other deduped warnings - document the convergence timeout escape hatch and the klient exclusion of enabledSystemPrompts * fix(agent-core-v2): dedupe the plugin budget warning and surface section read failures - emit plugin-sections-oversized once per skipped-plugin signature - let enabledSystemPrompts failures propagate to the refresh catch (keeps the current prompt and warns) instead of silently rendering and persisting a prompt without plugin instructions - cover the convergence timeout cut-off with a fake-timers test - clarify that the first-render timestamp anchors per process * fix(agent-core-v2): serialize session convergence and restore onDidReload timing - run at most one convergence per session and bound each change's wait by the timeout, so a fan-out emitter never interleaves deliveries after a timed-out convergence - fire onDidReload as soon as the reload commits again, keeping hook reloads independent of prompt convergence - sign the plugin budget warning with an unambiguous key * docs(agent-core-v2): align convergence wording with the serialized semantics - the timeout retry promise only holds once stalled work clears - note the per-session serial delivery cost model on the plugin change contract and the dual-queue invariant on the service * fix(agent-core-v2): keep empty plugin sections byte-neutral in the prompt template - place ${plugin_sections} on the same template line as ${skills_section} so prompts without either block render exactly as before this feature - note on the change contract that waitUntil work must not call back into plugin mutations, and spell out the per-session convergence order in the user docs * fix(agent-core-v2): pin a fork's profile so refresh triggers never rebind it - applyBindingSnapshot left the fork with no pinned profile, which routed in-process forks into the post-restart catalog rebind and could reset an inherited tool set; forks now inherit the source agent's pinned profile object - pin the first-render timestamp reuse with a ${now}-embedding test and document the anchored ${now} semantics - tighten the plugin docs budget and resume-refresh wording * fix(agent-core-v2): join in-flight convergence during agent bootstrap - an agent created while a plugin convergence is in flight now waits for it, and a restored agent refreshes once after it, so a plugin mutation never straddles an agent's bootstrap - warn on a non-string systemPrompt field and strip a UTF-8 BOM from systemPromptPath files before trimming - correct the consumption-surface wording (every CLI surface on the experimental flag, not just kimi -p), the per-session queueing note, and the single-plugin combined budget clause * fix(agent-core-v2): bound the bootstrap convergence join by the timeout A permanently wedged convergence kept convergeTail pending forever, and the unconditional settled() wait in bindBootstrap would have blocked every later agent creation in that session; the join now races the shared convergence timeout and continues (a restored agent still refreshes once, which never touches the tail), and the timeout constant moves to the contract for reuse * fix(agent-core-v2): close the convergence race against in-progress restores - a convergence fan-out could land while an agent's wire log is still replaying, dispatching a replay-visible config record whose effect the rest of the replay then overwrites; refreshSystemPrompt now skips while the wire restore is in progress - convergence completion is tracked by a generation counter; bootstrap compares it (after a bounded join) and refreshes a restored agent exactly once when a round completed after its creation began, replacing the wasConverging flag that could miss both windows * fix(agent-core-v2): bound each convergence so a wedged participant cannot stop the pipeline - the fan-out now races the convergence timeout, so convergeTail always settles: a permanently hung refresh delays its round (blocked entries drain oldest-first on later changes) instead of killing the session's convergence for good - warn when agent bootstrap stops waiting on a stalled convergence - diagnose a blank systemPromptPath and pin the plugin-root escape guard with traversal, absolute-path, and symlink tests * fix(agent-core-v2): bound the skill reload, preserve user-tool overlays, roll the prompt clock daily - the convergence's skill-reload segment now races the same timeout as the fan-out, so no segment of the pipeline can wedge a session for good; it continues with the previous catalog and retries next change - a cold rebind that resets the tool set replays session-added user tools onto the new base instead of dropping them for the rest of the process - the rendered timestamp re-anchors when the UTC date rolls over, so long-lived processes keep a fresh clock while steady-state renders stay byte-stable within a day - the plugin budget warning dedupes per plugin id, and the docs note that systemPromptPath content is frozen until the next reload * feat(agent-core-v2): converge cold plugin changes on resume through a drift-free gate - restore replays the persisted binding untouched, then bootstrap refreshes only when drift-free inputs changed while the session was cold: the catalog profile's tool set/denylist, or the plugin-sections baseline persisted alongside the prompt on the existing bind/update payloads; directory-listing and date drift wait for live triggers, so quiet resumes append no replay-visible records - the rendered timestamp is day-precision (UTC date at 00:00, re-anchored on rollover), keeping steady-state renders byte-stable across resumes and sessions on the same day - consolidate both timeout helpers onto a shared raceOutcome, and drop the generation counter the gate supersedes - align the plugin-sections precedence prose with the AGENTS.md disclaimer (no self-granted authority, system instructions win on conflict) * fix(agent-core-v2): bound the restored-prompt gate and land the sections baseline - the gate's plugin-sections read now races the convergence timeout, so agent creation never blocks behind an unrelated plugin mutation - refreshes serialize per agent through a tail, so overlapping triggers cannot write prompts out of order - when plugin sections change but a plugin-free custom prompt does not, the new baseline lands as a sections-only update instead of making every later resume re-render in vain - align the system prompt's Date and Time paragraph with the day-precision anchored timestamp * Update plugin system-prompt instructions in changeset Live sessions pick up plugin changes, while the default TUI and `kimi -p` paths ignore these fields. Signed-off-by: 7Sageer <sag77r@hotmail.com> * refactor(agent-core-v2): keep plugin skill reload user-driven Plugin mutations still converge live agent prompts, but the session skill catalog goes back to refreshing only on explicit plugin reload, as before: the prompt feature does not need skill convergence, and the pre-existing manual-reload semantics stay uniform across all plugin contributions. Removes the convergence-driven skill reload, the reloadSource de-privatization, and their tests; restores the PluginSkillSource onDidReload forwarding and its catalog tests. * refactor(agent-core-v2): apply plugin system-prompt changes only on explicit reload Drop the live convergence machinery (the plugin onDidChange barrier, the sessionPluginContribution fan-out, the restored-prompt drift gate, and the day-precision render clock) so plugin system-prompt sections take effect at the same point as every other plugin contribution: /plugins reload or a new session. The profile now refreshes when the session skill catalog re-pulls its plugin source on reload, reading both the skill list and the prompt sections fresh. * feat(agent-core): let plugins contribute system prompt instructions via the manifest systemPrompt field * chore(agent-core-v2): remove inline implementation comment * docs: clarify plugin prompt refresh semantics --------- Signed-off-by: 7Sageer <sag77r@hotmail.com> (cherry picked from commit 02d77b2) * feat: support plugin-contributed custom agents (MoonshotAI#2365) * feat: support plugin-contributed custom agents * fix: await plugin loading before agent catalog * fix: refresh plugin agents on v1 reload * test(agent-core-v2): add enabledSystemPrompts to the plugin service stub (cherry picked from commit fa2c5ce) * fix: remove the blocking wait from the TaskOutput tool (MoonshotAI#2379) * fix: remove the blocking wait from the TaskOutput tool The block/timeout parameters let a model stall the whole turn waiting for a background task (up to 3600s), even though completion already arrives via automatic notification. Remove both parameters from the v1 and v2 engines (kept in model-facing parity), simplify retrieval_status to success/not_ready, and update the tool, Bash, and Agent prompt wording plus user docs accordingly. Stale callers passing block are silently treated as a non-blocking snapshot. * fix: align background-task prompts with the non-blocking TaskOutput The compaction reminder promised TaskOutput could fetch a task's result for tasks that are still running, where it now returns not_ready — reword it to snapshot semantics and point at the completion notification. Also list AskUserQuestion(background=true) as a task source in the TaskOutput description. * test: exercise stale TaskOutput args through the runtime validator A stale block/timeout argument never reaches the tool: the executor's preflight validates args against the closed tool schema and rejects them immediately, so the old test documented silent-tolerance semantics the runtime never exhibits. Assert the real behavior through compileToolArgsValidator/validateToolArgs instead, and drop statement-adjacent comments to match the package's header-only comment convention. (cherry picked from commit 691ec46) * fix(vscode): all AskUserQuestions should be answered and added to the context (cherry picked from commit 3bbc3c5) * fix(vscode): open plan files from review (cherry picked from commit 729e69f) * fix(vscode): bind plan reviews to their files (cherry picked from commit f6b7f45) * fix(agent-core-v2): honor configured permission rules (cherry picked from commit ee0cc67) * fix(anthropic): omit incompatible tool schemas (cherry picked from commit 4e9ffbb) * fix(agent-core-v2): rescan agent profile catalog when a dispatch lookup misses Long-lived sessions (e.g. hosted by kimi web) scan agent file directories once at session materialization, so agent Markdown files written afterward can never be dispatched: the Agent tool, AgentSwarm tool, and swarm spawn path all fail with "Unknown agent type" until the server restarts. Add getProfileOrReload(), a small lookup helper in the sessionAgentProfileCatalog domain: on a miss it calls the catalog's existing (previously unwired) reload() once — deduped across concurrent misses — and retries before the caller's original error path applies. The three dispatch sites read the catalog live, so a rescan is visible to existing sessions immediately; the happy path is unchanged. (cherry picked from commit 27c4342) * fix(tui): refresh task output previews (cherry picked from commit 5d56451) * fix(tui): stop polling completed task output (cherry picked from commit 1a6df5b) * fix(tui): preserve final task output refresh (cherry picked from commit 56e7a72) * fix(tui): stop polling newly completed task output (cherry picked from commit ebe87b6) * fix(kimi-web): follow window width for the chat column The reading column was capped at 760px, leaving wide empty margins on large screens. Let it follow the pane width instead. (cherry picked from commit b6d88b5) * fix(kimi-web): swallow Ctrl+S in the composer to block the Save Page dialog preventDefault ran only while a turn was running, so an idle Ctrl+S/Cmd+S fell through to the browser and opened the Save Page dialog. Always swallow the shortcut; steering still only fires when a turn is running. (cherry picked from commit 7e44ece) * fix(kimi-web): anchor the conversation outline to the pane's right edge With the reading column no longer capped at a fixed width, the outline rail's old anchor (the 760px column edge) left it floating in the middle of the widened content. Pin the rail to the pane's right edge inside the scrollbar gutter, reveal labels leftward over the content on hover, and measure the expanded fit in that direction. (cherry picked from commit 6edc9c0) * fix(kimi-web): swallow Ctrl+S app-wide so the browser never opens Save Page The composer's keydown only fires while the composer is focused, so a Ctrl+S/Cmd+S pressed anywhere else still fell through to the browser's Save Page dialog. A capture-phase preventDefault in App.vue's global keydown now swallows the shortcut everywhere; steering stays in the composer's handler. (cherry picked from commit 0d9835e) * fix(kimi-web): keep code font metrics in bare markstream code blocks Plain-text code blocks rendered without the container wrapper nest <code> deeper than pre > code, so the inline-code chip rule (font: .9em) applied inside them and shrank the code to 10.8px/normal while the line-number overlay stayed at 12px/18px — the numbers drifted progressively below their lines. Extend the font:inherit guard to all markstream pres. (cherry picked from commit c1eab79) * fix(kimi-web): drop the TOC hover bridge and uncap the design-system demo The rail's invisible hover bridge sat over the right edge of full-width content and intercepted clicks and text selection from the messages beneath it; only the actual outline rows receive pointer events now. Also drop the 560px cap from the design-system chat demo so the spec matches the full-width behavior. (cherry picked from commit d78caaf) * Infer the Anthropic wire from a provider's /anthropic endpoint path The models.dev catalog resolver decided a provider's wire from type/npm/id only. A provider exposing its Anthropic-compatible surface under a /anthropic path, without an explicit type or an anthropic/claude token in npm/id, fell through to the OpenAI-compatible fallback and persisted the Anthropic endpoint under the OpenAI wire. Infer the Anthropic wire from a /anthropic endpoint path segment (matched on the path only), in both lockstep copies of the resolver, with unit and import-integration coverage. (cherry picked from commit 83e27aa) * fix(agent-core): nudge todo reconciliation when a turn ends with unfinished todos (cherry picked from commit aae8d7f) * fix(agent-core): honor same-turn TodoList writes in the turn-end reminder Addresses review feedback: the turn-end check now scans the whole just-finished turn (back to the user prompt that started it) for a TodoList write, so a mid-turn update followed by a closing text reply no longer triggers a spurious reminder. Also adds the required changeset. (cherry picked from commit 81f48ee) * fix(kap-server): defer global search activation (cherry picked from commit b514266) * fix(tui): preserve quotes in Windows status line commands (cherry picked from commit a555f1f) * test(tui): stabilize Windows quote regression (cherry picked from commit b7f549e) * fix(agent-core-v2): preserve select_tools for disclosure (cherry picked from commit a361352) * test(agent-core-v2): align optional harness option (cherry picked from commit bb3d4c8) * fix(tui): bound completed shell output frames (cherry picked from commit 6fb6bac) * feat(hooks): expose effective permission mode (cherry picked from commit 2dd7d01) * fix(mcp): reinitialize expired HTTP sessions (cherry picked from commit e03fd73) * fix(acp): stream agent-initiated turns (cherry picked from commit 0d7614c) * chore: prepare kkm 0.30.0 portable upstream release --------- Signed-off-by: 7Sageer <sag77r@hotmail.com> Co-authored-by: 7Sageer <sag77r@hotmail.com> Co-authored-by: wenhua020201-arch <wenhua020201@gmail.com> Co-authored-by: qer <wbxl2000@outlook.com> Co-authored-by: Kai <me@kaiyi.cool> Co-authored-by: rickgao <rickgao@tencent.com> Co-authored-by: zhaojy <zhaojy01@rd.netase.com> Co-authored-by: ericzhao.zhao <ericzhao.zhao@cyberklick.com> Co-authored-by: gunnlace <docdoby14470814@163.com> Co-authored-by: Ben Younes <benyounes.ousama@gmail.com> Co-authored-by: lostforwurdz <222525123+lostforwurdz@users.noreply.github.com> Co-authored-by: Tolik Trek <tolik.trek@gmail.com> Co-authored-by: octo-patch <266937838+octo-patch@users.noreply.github.com> Co-authored-by: airudotsh <airudotsh@users.noreply.github.com> Co-authored-by: Xule Lin <43122877+linxule@users.noreply.github.com> Co-authored-by: luren <lurenjia534@outlook.com> Co-authored-by: Codex <codex@openai.com>
7 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Related Issue
No linked issue — the problem is explained below.
Problem
Plugins currently have no native way to contribute instructions to the agent's system prompt. Existing manifest capabilities land elsewhere in the conversation:
sessionStartbecomes a user-role<system-reminder>injection,skillInstructionsis added only when one of the plugin's skills is activated, and hooks append user/assistant messages. The skills listing reaches the system prompt, but only as skill names and descriptions. A plugin that needs stable, always-on domain guidance therefore has no dedicated channel.What changed
This PR adds
systemPromptandsystemPromptPathtokimi.plugin.json:systemPromptcontains inline instructions.systemPromptPathpoints to a UTF-8 text file inside the plugin root.systemPromptPathis resolved while parsing the manifest and folded into the resolvedsystemPrompt, so the klient contract only needs the optional resolved string.Enabled contributions are exposed through
IPluginService.enabledSystemPrompts()and rendered with per-plugin provenance annotations plus the same precedence disclaimer used for workspace instruction files. The built-in prompts place a# Plugin Instructionsblock after the skills section. Custom v2 agent files andSYSTEM.mdtemplates can place the same block with${plugin_sections}.The parser limits each inline field and referenced file to 32 KB of UTF-8 content. Prompt assembly accepts at most 64 KB across enabled plugins; contributions beyond that aggregate budget are skipped with a warning. Unsafe paths, blank paths, unreadable files, non-string fields, and oversized content are reported through manifest diagnostics instead of escaping the plugin root or failing unrelated plugins.
Runtime behavior
Both agent engines consume the same manifest fields, diagnostics, provenance format, and budgets, but retain their existing lifecycle models:
kimi weband experimental CLI surfaces): a new Agent builds its prompt from the current App-scope plugin catalog. Explicit plugin reload commits a new catalog snapshot and emitsonDidReload; every live Session then re-pulls its plugin skill source, and after that refresh finishes its Agents request asynchronous prompt rebuilds from the current plugin sections. The v2reloadPlugins()call returns after the catalog commit and does not wait for those per-Session prompt rebuilds. Install, enable, disable, and remove update the App catalog but do not directly refresh existing Agent prompts; a later rebuild, such as compaction or a tool-policy/config refresh, reads the live catalog and may pick up those changes. A restored Agent first replays its persisted prompt unchanged and follows the same live-trigger behavior afterward.kimi -p): Session creation captures the current enabled plugin sections. Explicit plugin reload pushes the refreshed snapshot to every live Session and awaits prompt rebuilds for its ready Agents. Install, enable, disable, and remove without a reload leave live Session snapshots unchanged; new Sessions use the latest catalog.Requests already in flight keep the prompt snapshot with which they started. Toggling an individual plugin MCP server does not change plugin system-prompt sections.
Tests cover manifest parsing and path containment, diagnostics and byte budgets, plugin consumption reads, prompt composition and custom-template variables, v2 source-change prompt rebuild behavior, and legacy live-session refresh behavior. The user documentation covers the manifest fields and engine-specific behavior in English and Chinese.
Checklist
gen-changesetsskill, or this PR needs no changeset.gen-docsskill, or this PR needs no doc update.