Merge pull request #129 from ActiveMemory/feat/skill-updates

add /ctx-task-out — close the spec→implement gap in the canonical chain

Author: josealekhine

.claude/skills/ctx-experimental-handoff/SKILL.md

@@ -7,7 +7,7 @@ allowed-tools: Bash(specify:*), Read, Glob
 # Hand off an intent spec to spec-kit — experimental
 
 > **Experimental / discardable.** ctx-native port of an external
-> `anchor-handoff` skill. It is the delegation seam canonical ctx does
+> spec-handoff skill. It is the delegation seam canonical ctx does
 > not have: ctx's chain terminates at `/ctx-implement`, whereas this one
 > hands the intent spec to spec-kit. Trial it; if the seam is worth
 > keeping, promote it to a real `/ctx-handoff`.

.claude/skills/ctx-experimental-plan/SKILL.md

@@ -7,7 +7,7 @@ allowed-tools: Read, Write, Glob, Grep, Bash(date:*)
 # Plan scrutiny (adversarial interview) — experimental
 
 > **Experimental / discardable.** This is a ctx-native port of an
-> external `anchor-plan` skill, kept as a project-level skill so it can
+> external plan-scrutiny skill, kept as a project-level skill so it can
 > be trialed and deleted without touching ctx's canonical
 > `/ctx-plan`. It feeds the experimental spec-kit delegation chain
 > `/ctx-experimental-plan → /ctx-experimental-spec →

.claude/skills/ctx-experimental-spec/SKILL.md

@@ -7,7 +7,7 @@ allowed-tools: Read, Edit, Write, Glob, Grep
 # Write an intent spec from a debrief — experimental
 
 > **Experimental / discardable.** ctx-native port of an external
-> `anchor-spec` skill. It deliberately writes a **loose intent spec** in
+> intent-spec skill. It deliberately writes a **loose intent spec** in
 > a ctx-namespaced location (`.context/specs/`), separate from
 > spec-kit's repo-root `specs/<NNN-slug>/`, so the two trees never
 > collide. This differs from canonical `/ctx-spec`, which walks the full

.context/DECISIONS.md

@@ -3,6 +3,7 @@
 <!-- INDEX:START -->
 | Date | Decision |
 |----|--------|
+| 2026-07-04 | Statusline informs, never gates |
 | 2026-07-03 | Keep sonnet-4-6 at the 200k default despite the API catalog listing 1M |
 | 2026-06-07 | ctx-dream executor is a documented contract, not a hardcoded cron/claude assumption |
 | 2026-06-07 | Output belongs in write/ — taxonomy and emission style (consolidated) |
@@ -110,6 +111,20 @@ For significant decisions:
 
 -->
 
+## [2026-07-04-152957] Statusline informs, never gates
+
+**Status**: Accepted
+
+**Context**: Porting a cost-aware status line whose reference design escalates to a spend alarm with a model-switch suggestion when an expensive model family is detected
+
+**Decision**: Statusline informs, never gates
+
+**Rationale**: A family-substring rule carries no task context: the expensive model is often the cheaper choice per outcome, and a statusline cannot see outcomes, so it must not prescribe. Alarms also only reach operators who are present and already see the dollar figure; unattended overspend belongs to loop/cron notifications
+
+**Consequence**: ctx system statusline renders model, ctx%, and plain cost only; show_cost exists for screen-sharing; any future budget enforcement must be a separate, deliberate feature, not statusline creep
+
+---
+
 ## [2026-07-03-182236] Keep sonnet-4-6 at the 200k default despite the API catalog listing 1M
 
 **Status**: Accepted
@@ -656,7 +671,7 @@ that risk materializes.
 **Status**: Accepted
 
 **Context**: Per-session, operator-specific artifacts that grow without bound
-and can leak host/internal identifiers (ari, asgard, broadcom-class) into public
+and can leak host/internal identifiers into public
 mirrors when the project's .context/ is committed.
 
 **Decision**: Gitignore .context/handovers/; track only .gitkeep
@@ -686,8 +701,8 @@ already handles cross-machine knowledge persistence.
 **Decision**: Deprecate and remove ctx backup
 
 **Rationale**: Hub handles persistence, backup is env-specific, wrong layer for
-ctx to own. No external users depend on it. Broadcom mirror issue and GVFS
-Linux-only dependency add maintenance burden.
+ctx to own. No external users depend on it. An internal mirror issue and the
+GVFS Linux-only dependency add maintenance burden.
 
 **Consequence**: Need backup-strategy runbook before removal. Maintainer must
 set up replacement cron job. About 60 files to remove across CLI, config, hooks,

.context/LEARNINGS.md

@@ -17,6 +17,7 @@ DO NOT UPDATE FOR:
 <!-- INDEX:START -->
 | Date | Learning |
 |----|--------|
+| 2026-07-04 | Typed JSON round-trips silently drop user-owned keys |
 | 2026-07-03 | New Claude model families silently fall to the 200k default in ModelContextWindow |
 | 2026-06-07 | Pin an on-disk contract before splitting work across parallel agents |
 | 2026-06-07 | site/ is tracked build output — rebuild and bundle it with doc commits |
@@ -108,6 +109,16 @@ DO NOT UPDATE FOR:
 
 ---
 
+## [2026-07-04-152957] Typed JSON round-trips silently drop user-owned keys
+
+**Context**: The init permissions merge read settings.local.json into a typed struct and re-marshaled it, so every key ctx does not model (env, a user's statusLine) vanished on re-init
+
+**Lesson**: A typed read-modify-write of a file users also edit by hand is silent data loss; unknown fields do not survive encoding/json round-trips
+
+**Application**: Mutate shared JSON via raw-map surgery (map[string]json.RawMessage), rewriting only the keys ctx owns; see internal/cli/initialize/core/merge/settings.go
+
+---
+
 ## [2026-07-03-182238] New Claude model families silently fall to the 200k default in ModelContextWindow
 
 **Context**: claude-fable-5 sessions warned '104 percent full' while the 1M window was 79 percent free; ModelContextWindow only recognized the [1m] suffix and the opus substring

.context/TASKS.md

@@ -2429,7 +2429,7 @@ first
 the
 > serendipity nag. The `dream-guard` consolidation is the second open item.
 > Baseline is intentionally being frozen here for a controlled snapshot-VM
-> experiment (anchor SDD / spec-kit alignment); do not assume this phase is
+> experiment (SDD / spec-kit alignment); do not assume this phase is
 shipped.
 
 - [ ] Close the end-user dream UX loop: invoking /ctx-dream interactively is a
@@ -2497,6 +2497,16 @@ shipped.
 
 ### Misc
 
+- [x] Implement ctx system statusline per specs/statusline.md: stdin-JSON render (model, ctx%, cost), .ctxrc statusline block, setup merge with backup/restore. Informational only; no gating (spec records why) #priority:medium #session:a31b3e67 #branch:main #commit:687bbd59 #added:2026-07-04-140249
+
+- [x] Companion-tool feature-delta analysis; borrow/skip verdicts recorded in gitignored inbox/ notes (local only) #session:a31b3e67 #branch:main #commit:687bbd59 #added:2026-07-04-135227
+
+- [x] Add ctx-humanize plugin skill (SKILL.md + references/pattern-catalog.md, docs entry) with progressive disclosure and voice guardrails. Spec: specs/ctx-humanize.md #session:a31b3e67 #branch:main #commit:687bbd59 #added:2026-07-04-134612
+
+- [x] Add hack/check-tools.sh tooling dependency checker (manifest: hack/tool-versions.txt, make check-tools) — spec: specs/check-tools.md #session:a31b3e67 #branch:main #commit:687bbd59 #added:2026-07-04-132852
+
+- [ ] Recipe: the full design-to-implementation pipeline from the operator's seat (new 'spec-driven-development.md' or a major fold into design-before-coding.md). Must cover, for a newcomer with zero tribal knowledge: (1) the 5-step chain INCLUDING /ctx-plan — design-before-coding.md's TL;DR currently omits the debated-brief step entirely; (2) altitude: the bet is debated once and the spec covers ALL milestones — briefs are per-bet, never per-milestone; (3) plans are just-in-time per milestone behind the rolling-wave gate (tasking distant milestones produces fiction); (4) blocking-TBD gates as the replacement for per-milestone debates — each task-out run forces exactly the decisions that milestone embeds, into DECISIONS.md; (5) two surfaces, one truth: plan = execution ledger (st column), TASKS.md epics = one-way projections over disjoint id ranges; DoD is measurement/Board-confirmed, never derived; (6) when a NEW brief happens: new bet (e.g. deferred machinery returning) or evidence falsifying the committed bet — never relitigating from below; (7) a worked multi-milestone example, not a one-session feature. Origin: zhc/os session a63353a3 (2026-07-03) — the operator had to reverse-engineer all seven from skill texts and agent explanations #priority:high #session:a63353a3 #branch:main #commit:511a609a #added:2026-07-03-232251
+
 - [ ] Tighten /ctx-spec (and /ctx-implement) skills to bake in spec-kit's one
   good discipline without adopting spec-kit: every /ctx-spec must end with (a)
   explicit, testable acceptance criteria and (b) a derived, dependency-ordered
@@ -2507,3 +2517,11 @@ shipped.
   memory + adversarial /ctx-plan; avoid two-constitutions double-bookkeeping).
   #priority:medium #session:210b77dd #branch:main #commit:6b0d0107
   #added:2026-06-27-222130
+
+### Future
+
+- [ ] Hub curation: immutable promotion ledger (who accepted what, when, why) + mechanical validate floor for shared knowledge; revisit when ctx hub grows team-curation workflows #priority:low #session:a31b3e67 #branch:main #commit:d800734c #added:2026-07-04-153004
+
+- [ ] ctx-spec-views skill: manager-facing read-models (execution plan, spec briefs, task breakdowns) generated FROM specs/, never source of truth; spec it when someone actually needs the leadership view #priority:low #session:a31b3e67 #branch:main #commit:d800734c #added:2026-07-04-153004
+
+- [ ] KB convention: pinned upstream corpus for grounding — document a snapshot mode (dated local copy of high-churn upstream docs as the citable byte-stream) in KB rules; no code needed #priority:low #session:a31b3e67 #branch:main #commit:d800734c #added:2026-07-04-153004

Makefile

@@ -5,7 +5,7 @@
 .PHONY: build ctxctl test vet fmt fmt-context lint lint-style lint-drift lint-shellcheck lint-powershell \
 clean all release build-all help \
 test-coverage smoke site site-feed site-serve site-serve-lan site-setup audit check plugin-reload \
-journal journal-serve journal-serve-lan gpg-fix gpg-test register-mcp reinstall \
+journal journal-serve journal-serve-lan gpg-fix gpg-test register-mcp reinstall check-tools \
 sync-version check-version-sync sync-why check-why sync-copilot-skills check-copilot-skills sync-steering check-steering gemini-search \
 gitnexus-version gitnexus-update install-ctxctl reinstall-ctxctl
 
@@ -276,6 +276,10 @@ gpg-fix:
 gpg-test:
 	./hack/gpg-fix.sh --test
 
+## check-tools: Verify tooling dependency versions (Go, Node, GitNexus, MCP servers, ...)
+check-tools:
+	@./hack/check-tools.sh
+
 ## register-mcp: Register all MCP servers (gemini-search, gitnexus) with Claude Code
 register-mcp:
 	@./hack/register-gemini-search.sh

README.md

@@ -19,7 +19,7 @@ vendor lock-in.
 `ctx` is infrastructure for preserving intent under scale. Without durable
 context, intelligence resets. With `ctx`, creation compounds.
 
-Read the full **[Manifesto](MANIFESTO.md)** | **[ctx.ist/manifesto](https://ctx.ist/manifesto/)**
+Read the full **[Manifesto](MANIFESTO.md)** | **[ctx.ist](https://ctx.ist/)**
 
 ## The Thesis
 
@@ -36,7 +36,7 @@ Read the full **[Thesis](https://ctx.ist/thesis/)**
 
 | Document                                                          | Context                                             |
 |-------------------------------------------------------------------|-----------------------------------------------------|
-| [Manifesto](https://ctx.ist/manifesto/)                           | Philosophy: creation, context, verification         |
+| [Manifesto](https://ctx.ist/)                                     | Philosophy: creation, context, verification         |
 | [The Thesis](https://ctx.ist/thesis/)                             | Whitepaper: context as deterministic state          |
 | [Design Invariants](https://ctx.ist/reference/design-invariants/) | System properties that must always hold             |
 | [Tool Comparison](https://ctx.ist/reference/comparison/)          | How `ctx` differs from .cursorrules, Aider, Copilot |
@@ -165,9 +165,9 @@ recipes are the right next stop.
 |-------------------------------------------------|----------------------------------------|
 | [Getting Started](https://ctx.ist)              | Installation, quick start, first steps |
 | [Recipes](https://ctx.ist/recipes/)             | Practical workflow guides              |
-| [CLI Reference](https://ctx.ist/cli-reference/) | All commands and options               |
-| [Context Files](https://ctx.ist/context-files/) | File formats and structure             |
-| [Integrations](https://ctx.ist/integrations/)   | Claude Code, Cursor, Aider setup       |
+| [CLI Reference](https://ctx.ist/cli/)           | All commands and options               |
+| [Context Files](https://ctx.ist/home/context-files/) | File formats and structure       |
+| [Integrations](https://ctx.ist/operations/integrations/) | Claude Code, Cursor, Aider setup |
 | [Operations](https://ctx.ist/operations/)       | Runbooks, day-to-day, hub deployment   |
 | [Security](https://ctx.ist/security/)           | Trust model, audit trail, permissions  |
 

docs/blog/2026-01-27-building-ctx-using-ctx.md

@@ -82,7 +82,7 @@ The core idea was simple:
 [**Ralph Loop**][ralph] (*An iterative AI development workflow*)
 had produced a working CLI:
 
-[ralph]: https://ctx.ist/autonomous-loop/ "Autonomous Loop"
+[ralph]: https://ctx.ist/operations/autonomous-loop/ "Autonomous Loop"
 
 ```
 feat(cli): implement amem init command

docs/blog/2026-02-01-ctx-v0.2.0-the-archaeology-release.md

@@ -264,7 +264,7 @@ The same structure serves two very different readers.
 ## The Configuration: `.contextrc`
 
 [Projects can now customize `ctx` behavior 
-via `.contextrc`](../cli/system.md#contextrc).
+via `.contextrc`](../home/configuration.md).
 
 This makes `ctx` usable in real teams, not just personal projects.