DocMason
CLAUDE.md152 starsChanged 4 months agoApache-2.0View on GitHub
This is a DocMason workspace. Read the full agent contract below. @../AGENTS.md This file is the committed Claude Code entry surface for this repo. AGENTS.md is the baseline contract. Nothing below overrides AGENTS.md; it only gives Claude-specific entry guidance for the same contract. For an ordinary user request on Claude Code: For explicit operator work: Bootstrap, repair, or docmason sync-adapters may generate repo-local Claude helpers such as: Treat those as generated local helpers, not authored instructions. Do not hand-edit them,…
In this repository17 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- adapter-syncSkill
- askSkill
- grounded-answerSkill
- grounded-compositionSkill
- knowledge-base-syncSkill
- knowledge-constructionSkill
- provenance-traceSkill
- retrieval-workflowSkill
- runtime-log-reviewSkill
- validation-repairSkill
- workspace-bootstrapSkill
- workspace-doctorSkill
- workspace-statusSkill
- operator-evalSkill
- public-sample-workspaceSkill
The file
Show its rules and text
We split this CLAUDE.md into the rules it gives an agent
15 rules. About 0 are common, 2 are distinctive, and 13 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
13
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 15
after AGENTS.md, the normal next read is skills/canonical/ask/SKILL.md
- Very specific: line 17
do not start with workspace-status, workspace-bootstrap, source search, CLI help, or ad hoc implementation reading unless the request is …
- Very specific: line 21
follow the matching canonical workflow or stable docmason CLI from AGENTS.md
- Distinctive: line 22
do not force setup, status, sync, or review work through the ordinary ask front door
- Very specific: line 29
Treat those as generated local helpers, not authored instructions.
- Very specific: line 30
Do not hand-edit them, create global skill links, or maintain parallel skill copies.
- Very specific: line 36
When .venv is absent, the wrappers exit quietly and ordinary bootstrap guidance still comes from AGENTS.md and the canonical skills.
- Very specific: line 38
Do not treat low-level Python helpers in docmason.ask as preferred host entrypoints.
- Very specific: line 42
If generated Claude helpers exist, use them to avoid reverse engineering the repo.
- Very specific: line 43
If they do not exist yet, continue directly from AGENTS.md and skills/canonical/ask/SKILL.md.
- Very specific: line 44
Use committed hooks only as optional host plumbing, not as the ordinary ask front door.
- Very specific: line 45
Do not reverse engineer ask.py or substitute retrieve / trace for canonical ask completion.
- Distinctive: line 46
Do not return a final business answer unless the canonical turn has already reached legal completion or governed boundary closure.
- Very specific: line 49
If the file exists, Claude gets the richer routing layer.
- Very specific: line 50
If it does not exist yet, continue from AGENTS.md and the canonical ask contract rather than improvising a host entry path.
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
DocMason
This is a DocMason workspace. Read the full agent contract below.
@../AGENTS.md
Claude Code Notes
This file is the committed Claude Code entry surface for this repo.
AGENTS.md is the baseline contract.
Nothing below overrides AGENTS.md; it only gives Claude-specific entry guidance for the same contract.
For an ordinary user request on Claude Code:
- after
AGENTS.md, the normal next read isskills/canonical/ask/SKILL.md - then open canonical
askthrough theCanonical Ask Contractin that skill and follow the returned route - do not start with
workspace-status,workspace-bootstrap, source search, CLI help, or ad hoc implementation reading unless the request is clearly explicit operator work or canonicalaskreturns that route
For explicit operator work:
- follow the matching canonical workflow or stable
docmasonCLI fromAGENTS.md - do not force setup, status, sync, or review work through the ordinary ask front door
Bootstrap, repair, or docmason sync-adapters may generate repo-local Claude helpers such as:
adapters/claude/project-memory.mdfor richer workflow routing.claude/skills/for repo-local Claude skill discovery
Treat those as generated local helpers, not authored instructions.
Do not hand-edit them, create global skill links, or maintain parallel skill copies.
The committed .claude/settings.json plus .claude/hooks/ are the repo's Claude host plumbing.
They provide best-effort native capture, minimal continuation context, one-shot closure, and
repo-local shim refresh; they never own canonical ordinary ask execution.
Claude Code may skip them until folder trust is accepted, when Hooks are disabled, or when managed
policy permits only managed Hooks. That never changes the canonical workflow contract.
When .venv is absent, the wrappers exit quietly and ordinary bootstrap guidance still comes from
AGENTS.md and the canonical skills.
Do not treat low-level Python helpers in docmason.ask as preferred host entrypoints.
They are internal lifecycle primitives behind the canonical ask workflow contract.
Generated helpers and adapter routing are derived convenience surfaces for Claude Code.
These generated helpers should help Claude follow the authored canonical contract, not replace it.
If generated Claude helpers exist, use them to avoid reverse engineering the repo.
If they do not exist yet, continue directly from AGENTS.md and skills/canonical/ask/SKILL.md.
Use committed hooks only as optional host plumbing, not as the ordinary ask front door.
Do not reverse engineer ask.py or substitute retrieve / trace for canonical ask completion.
Do not return a final business answer unless the canonical turn has already reached legal completion or governed boundary closure.
The import below is optional enrichment.
If the file exists, Claude gets the richer routing layer.
If it does not exist yet, continue from AGENTS.md and the canonical ask contract rather than improvising a host entry path.
@../adapters/claude/project-memory.md
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.