skills: give every description a one-line lead the skill list can show in full #77

Merged

vnprc merged 1 commit from agent/skill-description-lead into master

2026-10-01 00:36:01 +01:00

allod-agent commented

2026-10-01 00:29:38 +01:00

Member

Closes allod/memory#76
Refs allod/tools#259

What

Reformat the six public skills' frontmatter descriptions so the first sentence stands alone in at most 100 characters — the shape allod skill's list line can show in full. The "Use when …" trigger material stays in the description tail; nothing was dropped, only reordered. The skills index in README.md now states the shape so new skills inherit it.

Leads (all measured under 100 characters):

delegate-implementation — "Split an implementation arc across agents: one overseer, workers in worktrees, cold reviewers." (94)
fleet-diff — "Gate a merge against which machines it actually rebuilds." (57)
forge-groom — "Triage every open issue and pull request in the allod repos against what has actually landed." (93)
list-models — "Find out which models each installed agent CLI declares and whether a named one answers." (88)
memory-backpass — "Read recent session traces against the memory indexes and open one PR of evidence-backed edits." (95)
pick-next-issue — "Sweep every open issue in the repos this agent can push to and pick the next one to implement." (94)

Risk

Low: frontmatter descriptions only; no skill body changed. The list-models and forge-groom tails were lightly reworded while reordering ("run it before scripting anything around a model name", "one or more repos at a time", "on master") — material preserved, but worth a skim to confirm the meaning survived. The README's 100-character figure mirrors briefLimit in allod/tools and could drift if that constant ever moves.

Validation

Lead lengths measured: every first sentence is at most 94 characters.
Each rewritten description carries its original "Use when" material in the tail; the diff shows reordering, no deletions.
The pre-commit hooks (memory-budget, pre-commit) pass; README.md and skills/ are exempt surfaces for the byte budget.

Closes allod/memory#76 Refs allod/tools#259 ## What Reformat the six public skills' frontmatter descriptions so the first sentence stands alone in at most 100 characters — the shape `allod skill`'s list line can show in full. The "Use when …" trigger material stays in the description tail; nothing was dropped, only reordered. The skills index in README.md now states the shape so new skills inherit it. Leads (all measured under 100 characters): - delegate-implementation — "Split an implementation arc across agents: one overseer, workers in worktrees, cold reviewers." (94) - fleet-diff — "Gate a merge against which machines it actually rebuilds." (57) - forge-groom — "Triage every open issue and pull request in the allod repos against what has actually landed." (93) - list-models — "Find out which models each installed agent CLI declares and whether a named one answers." (88) - memory-backpass — "Read recent session traces against the memory indexes and open one PR of evidence-backed edits." (95) - pick-next-issue — "Sweep every open issue in the repos this agent can push to and pick the next one to implement." (94) ## Risk Low: frontmatter descriptions only; no skill body changed. The list-models and forge-groom tails were lightly reworded while reordering ("run it before scripting anything around a model name", "one or more repos at a time", "on master") — material preserved, but worth a skim to confirm the meaning survived. The README's 100-character figure mirrors `briefLimit` in allod/tools and could drift if that constant ever moves. ## Validation - Lead lengths measured: every first sentence is at most 94 characters. - Each rewritten description carries its original "Use when" material in the tail; the diff shows reordering, no deletions. - The pre-commit hooks (memory-budget, pre-commit) pass; README.md and skills/ are exempt surfaces for the byte budget.

allod-agent added 1 commit

2026-10-01 00:29:38 +01:00

skills: every description opens with a one-line lead under 100 characters 924221347f

The skill list truncates at 100 characters, and every public description
ran past it, cutting off mid-identity. Each description now opens with a
first sentence that stands alone as the skill's identity; the Use when
triggers and detail follow unchanged in the tail. README's skills index
states the shape so new skills inherit it.