shaman
Skill1 starsChanged 52 days agoApache-2.0View on GitHub
Have the AI run a read-only orchraft setup preflight. Use when the user says shaman, asks whether orchraft is ready, or wants a preflight before another workflow.
The file
Show its rules and text
We split this skill into the rules it gives an agent
14 rules. About 0 are common, 7 are distinctive, and 7 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Use this skill before ship, roast, or land when setup is uncertain.
- Distinctive: line 14
Run the checks below without changing files, git state, GitHub, or Trello.
- Very specific: line 36
If a local policy exists, read it as the target repository's authority, but verify that context/policies/approval-policy.md matches the …
- Distinctive: line 41
When checking Autonomous Mode from the active config, apply the shared config-policy verification rules: compare committed config with the …
- Very specific: line 49
If gh exists, run gh auth status and report its result without exposing tokens.
- Distinctive: line 54
Do not treat a successful local git command as proof of GitHub authentication.
- Very specific: line 67
Check .orchraft.jsonc first, then .orchraft.json, and ignore both when neither exists because config is optional.
- Very specific: line 80
If hooks/hooks.json is present, validate it as JSON, confirm every command it references exists, and confirm referenced scripts are …
- Distinctive: line 88
Use this compact format:
- Very specific: line 103
Use FAIL for a broken prerequisite or malformed config.
- Distinctive: line 112
Read-only: never create or update a branch, commit, push, fetch, open or update a PR, change an issue or release, mutate Trello, or edit …
- Distinctive: line 114
Do not run commands that write caches, generated files, credentials, or dependency locks.
- Distinctive: line 116
Do not print secrets, tokens, full credential paths, or config values that may contain secrets.
- Very specific: line 118
Stop after the report. The shaman may recommend warplan, forgehand, ship, roast, or land, but it never invokes the next workflow.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill 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
Shaman Workflow
Use this skill before ship, roast, or land when setup is uncertain. It
reads the signs of the current repository; it never repairs anything.
At A Glance
- Resolve the current repository root and the orchraft plugin root.
- Run the checks below without changing files, git state, GitHub, or Trello.
- Report each check as
PASS,WARN, orFAIL, then give the smallest next action and an explicitREADYorNOT READYresult.
Checks
Repository and skill loading
- Confirm
git rev-parse --show-toplevelsucceeds and report the repository root and current branch. - Confirm the loaded orchraft package contains
skills/and the skill files named by the requested workflow. If this skill is running, its own loading is a pass; inspect the package for the other requested skills rather than claiming every skill is available. - Confirm the repository has an
originremote when a GitHub workflow is requested. Do not fetch, push, or alter remotes.
Policy fallback
- For
approval-policy.md,config-policy.md, andverification-policy.md, report whether the target repository provides a local copy or orchraft will use the bundled fallback. - If a local policy exists, read it as the target repository's authority, but
verify that
context/policies/approval-policy.mdmatches the freshly resolved default-branch copy before honoring itsAutonomous Mode. A missing, unreadable, or mismatched copy cannot pre-authorize actions. A missing local policy is a warning only; fallback is supported. - When checking Autonomous Mode from the active config, apply the shared config-policy verification rules: compare committed config with the default branch, or require a trusted exact-path ignore source for a local config absent from that branch. Failed or mismatched verification cannot authorize actions.
GitHub access
- If
ghexists, rungh auth statusand report its result without exposing tokens. A missing or unauthenticatedghis a warning when the configured GitHub connector can provide read access, otherwise a failure for workflows that need GitHub state. - Confirm the origin URL is a GitHub repository when GitHub work is requested. Do not treat a successful local git command as proof of GitHub authentication.
Required tools
gitis required for every workflow.jqis required for the bundled hooks; missingjqis a warning that ambient nudges will be silent.ghis required only when no authenticated GitHub connector is available.- Report missing optional tools plainly instead of inventing a workaround.
Config parsing
- Check
.orchraft.jsoncfirst, then.orchraft.json, and ignore both when neither exists because config is optional. - Parse the selected file as JSON. For
.jsonc, use the same quote-aware comment stripping convention ashooks/watchtower-nudge.sh; never strip//inside a quoted URL. - A malformed file is a
WARNthat names the file and parse error, then explain that workflows will continue with documented defaults. Do not edit or rewrite the file. - For a valid file, check only the documented types that affect the requested workflow. Unknown keys are not failures.
Hook support
- If
hooks/hooks.jsonis present, validate it as JSON, confirm every command it references exists, and confirm referenced scripts are executable. - Confirm
jqis available for the bundled hook scripts. - Report unsupported host hook behavior as a warning with the host-specific limitation. Never install, enable, or modify a hook.
Output
Use this compact format:
Repository: <root> (<branch>)
<status> Skills: <result>
<status> Policies: <local/fallback result>
<status> GitHub: <authenticated / connector / unavailable>
<status> Tools: <result>
<status> Config: <absent / valid path>
<status> Hooks: <supported / not installed>
Next action: <smallest action, or none>
Result: READY | NOT READYUse FAIL for a broken prerequisite or malformed config; any FAIL in a
requested workflow means NOT READY. Use WARN for a supported fallback or
unavailable optional integration; a warning permits READY only when the
requested outcome still works through that fallback or the warning concerns
an unavailable optional integration. Use PASS only for a check actually
performed in this turn. Never claim a check passed from memory.
Guardrails
- Read-only: never create or update a branch, commit, push, fetch, open or update a PR, change an issue or release, mutate Trello, or edit files.
- Do not run commands that write caches, generated files, credentials, or dependency locks.
- Do not print secrets, tokens, full credential paths, or config values that may contain secrets.
- Stop after the report. The shaman may recommend
warplan,forgehand,ship,roast, orland, but it never invokes the next workflow.
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.

