alexion/skills
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(agents): make skill metadata agent-agnostic

Browse Source
This commit is contained in:
alexion
2026-07-29 22:39:02 -04:00
parent 40b37281d5
commit b541251f98
21 changed files with 131 additions and 80 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
.claude/adr Symbolic link
View File

@@ -0,0 +1 @@
../.agents/adr

17
.claude/adr/0001-consume-writes-target-push-channel.md
View File

@@ -1,17 +0,0 @@
# Consume writes the target's push channel, not only the vault
`consume` was scoped to write only inside the vault's `01 sources/claude/` area.
A real run showed that most mined knowledge is not wiki-shaped — specific residue and proactive rules belong in the target's own always-loaded `CLAUDE.md` — and dropping it lost real signal, while redoing the target-doc edits by hand after each run was toil.
We decided `consume` writes a second channel outside the vault, the target's `CLAUDE.md`, rather than only planning those edits for the user to apply.
## Considered Options
- **Plan-only** — spell the target-doc edits out in the report and have the user apply them by hand.
Rejected: the whole value is not redoing what the run already worked out.
- **Auto-write both channels (chosen)** — apply the target edits too, but present them as a distinct, independently-approvable write-set under the one plan-approval gate, so a write outside the vault is never a surprise.
## Consequences
`consume`'s blast radius now includes files outside the vault and outside the spec/task scaffolding it deletes.
The plan-approval gate is what keeps that safe: the target write-set is approved separately from the vault write-set, and a declined target write-set degrades to flagged suggestions.
Destructive edits to `CLAUDE.md` are gated on capture — an entry is removed or trimmed only when its general part is present in the wiki.