Skip to content

Add agent merge bar note - #3

Merged
cschanhniem merged 1 commit into
mainfrom
agent-merge-bar
Jun 7, 2026
Merged

cschanhniem merged 1 commit into
mainfrom
agent-merge-bar

Conversation

@cschanhniem

Copy link
Copy Markdown
Contributor

Adds docs/AGENT-MERGE-BAR.md with a short note on agent contribution quality.

Validation not run locally in this connector session.

@cschanhniem
cschanhniem merged commit d4b2034 into main Jun 7, 2026
1 check passed
cschanhniem added a commit that referenced this pull request Jun 28, 2026
The repo's thesis is 'CI/CD for truth claims': make false AI-generated
knowledge harder to merge. But the protocol had no persistent claim
object — only transient submissions and single-source evidence records.

This commit adds the missing core objects:

- schemas/claim.schema.json: a persistent, falsifiable statement with a
  verification lifecycle (unverified -> dry-lab-verified -> replicated ->
  accepted -> field-tested), evidence links, failure modes, a kill
  condition, and required reviewer roles. High-safety claims cannot
  advance without red-team review. Falsified claims stay in the record
  so contributors do not repeat them.

- schemas/replication.schema.json: an independent replication record
  with replicator identity, environment, input hash, result, and
  divergence tracking. Self-replication is not replication.

Evidence schema strengthened:
- Added model-prediction, computational-analysis, and wet-lab-confirmation
  evidence types. The thesis envisions computational claims (molecule
  candidates, theorem sketches, benchmark results) but the enum had no
  way to represent them.
- Added optional doi and archive_url fields to insulate records against
  link rot.

Agent-submission schema strengthened:
- Added kill_condition as a required field. SKILL.md pattern #3 already
  required it in prose; the schema now enforces it. Protocol drift fixed.

Validator updated to compile and validate all new schemas and examples.

Stale pack counts fixed: AGENTS.md had a 21-pack table; ROADMAP.md said
15. Actual count is 100. Replaced with pointers to generated indexes,
following the repo's own rule: 'Never trust an embedded count in any
hand-written file.'

AI-slop language cleaned: replaced vague metaphorical 'landscape' in the
skills-training pack with precise terms. Legitimate ecological uses of
'landscape' (landscape simplification, agricultural landscapes) in
biodiversity packs were correctly left intact.

All validation passes: pnpm validate, pnpm reproducibility:check,
pnpm build.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant