| name | skillroster |
| description | Route tasks to specialized local Core and On-demand Skills, and govern the local Skill roster. Before reading or changing the workspace for a task that may need instructions not already visible, preserve the complete user message verbatim as TASK. For non-English or mixed input, run `skillroster find --hint "ONE FAITHFUL ENGLISH CAPABILITY PARAPHRASE" --load --limit 1 --json -- "TASK"`; for English input omit `--hint`. Follow the complete verified SKILL.md in `result.loaded_skill.content.text`. Also use for inventory, usage evidence, duplicates, broken links, Core/On-demand recommendations, Plans, Receipts, Apply, and Undo. Not for installing third-party Skills or migrating, distributing, synchronizing, or repairing shared Skill-manager directories.
|
| metadata | {"bootstrap-version":"1.8.29","skillroster-routing-triggers":"route task to local Skill; inventory installed Agent Skills; analyze duplicate or unused Agent Skills; govern a Skill Roster; prepare or apply approved Skill Plan; create or undo Skill Receipt"} |
SkillRoster
Use the local skillroster binary as the deterministic source of facts. Every
command uses explicit --json. Validate schema_version and ok before using
result. Typed suggested_actions are options, never authorization.
Choose one path
- For SkillRoster inventory, evidence, governance, setup, Apply, Undo, lifecycle,
or recovery requests, skip the Route gate and read the matching reference
below. Do not route SkillRoster back to itself.
- For another task, follow an already-visible exact Skill when one clearly
applies. Otherwise complete the Route gate.
Route gate
The Find call below must be the first task tool call. Before it, do not read or
change the workspace and do not execute a non-routing command:
- Keep the complete user message as
TASK, byte-for-byte in its original
language, including paths, limits, and output requirements. Do not summarize,
translate, shorten, or quote only part of it.
- If
TASK is non-English or mixed-language, create one concise English
capability paraphrase as HINT. It supplements TASK; it never replaces it.
- Invoke the fixed SkillRoster executable with
TASK and HINT as separate,
literal argv values; never interpolate either into shell syntax. Use:
- English:
skillroster find --load --limit 1 --json -- "TASK"
- Otherwise:
skillroster find --hint "HINT" --load --limit 1 --json -- "TASK"
The -- boundary is required even when TASK does not begin with -; it
keeps opaque user text from being parsed as a SkillRoster option.
- When
HINT was used, first require
ranking_strategy: task_hint_reciprocal_rank_fusion.
- Require
loaded_skill.selection.rank: 1, content.complete: true, and all
verification identity/digest checks to be true. The complete instructions
are loaded_skill.content.text; no second filesystem, workspace, MCP, or
SkillRoster read is needed. For a wrong-domain result or typed load blocker,
read references/routing.md and follow its bounded branch.
- Follow the loaded instructions and perform the original task. Treat
task_success: not_evaluated literally; only the task's own evidence can
establish success.
If the result is empty or clearly from another domain, retry once. Keep TASK
unchanged; refine the existing hint, or add one capability hint when the first
English call had none. If no usable result remains, stop and report a routing
failure.
Govern a roster
For inventory, usage, exposure, duplicate, broken-link, Core/On-demand, or Plan
requests, read references/governance.md before acting. Inspection and planning
do not change Agent files. Prefer the bounded summary first; load exhaustive
detail only for a decision that needs it.
Mutate or recover
For setup, Apply, Undo, lifecycle cleanup, or recovery, read
references/mutation.md before acting. A validated Plan is not permission to
apply it. Show the complete bounded impact and obtain one explicit confirmation.
Only a successful Receipt verifies that Agent files changed.
Never parse styled terminal output, invent a health score, infer token savings,
claim unobserved usage, or weaken ambiguity, drift, trust, and recovery checks.
In the final response, state whether files changed; claim changes only from a
verified Receipt or from task tools actually used after routing.