skill-prune

Name: skill-prune
Author: transilienceai

423stars74forks

How to install skill-prune?

transilienceai/communitytools/skill-prune

$npx -y skills add transilienceai/communitytools --skill skill-prune

Installs into the current project.

Summary

Identify and remove negative-ROI skill content — orphan files, never-read entries, duplicates, content reintroducing challenge-specific lore. Inverse of /skill-update. Use during quarterly maintenance or after the linter flags issues.

Files · 1

View on GitHub

SKILL.md

1	# Skill Prune
2
3	Inverse of `/skill-update`. Removes content rather than adding it. Run during quarterly maintenance, after engagements, or when `scripts/skill_linter.py` reports orphans / duplicates.
4
5	## When to invoke
6
7	- Quarterly cadence.
8	- After `scripts/skill_linter.py --check-orphans` reports orphan reference files.
9	- After a SKILL.md or reference file grows past its cap and needs trimming.
10	- After a de-specialization sweep, to drop content tied to retired challenges.
11
12	## Prune criteria (the four signals — same shape as /skill-update, inverted)
13
14	A reference / scenario / line is a prune candidate when it satisfies any of:
15
16	1. Orphan — not linked from any SKILL.md or other reference file in the last 60 days.
17	2. Referenced only by failed engagements — appeared in `attack-chain.md` of runs that ended `status=BLOCKED`, never in a successful chain.
18	3. Contradicted by newer content — a later scenario / pattern supersedes it; the older entry no longer reflects current technique.
19	4. Redundant with newer content — same technique covered more clearly elsewhere.
20
21	Removing content fails any of these → keep it.
22
23	## Safety rules
24
25	- Never prune a file with `<!-- KEEP: <reason> -->` annotation.
26	- Never prune content cited in a still-open engagement's `OUTPUT_DIR/attack-chain.md`.
27	- Never prune the canonical-home file for a single-owner rule (brute-force, output-discipline, env-reader, skill-update).
28	- Bias toward keeping technique-rich content over operational lore.
29
30	## Procedure
31
32	1. Run `scripts/skill_linter.py --check-orphans` to surface orphans.
33	2. For each candidate file or block, evaluate the four signals.
34	3. Build a deletion plan — show files / lines to remove with one-line rationale per item.
35	4. Apply deletions only after the plan is approved (skill-prune does not auto-delete during invocation).
36	5. Re-run `scripts/skill_linter.py` to confirm the change broke no other links and didn't reintroduce duplicates.
37
38	## Output
39
40	Concise change report:
41
42	- Removed. File or block + one-line rationale per item.
43	- Kept (flagged for follow-up). Items that failed all four signals but are worth revisiting next quarter.
44	- No prunes. State explicitly when nothing warranted removal.
45
46	## Anti-Patterns
47
48	- Pruning content because it's old — age alone is not a signal; relevance is.
49	- Pruning a reference because its parent SKILL.md is bloated — fix the SKILL.md instead.
50	- Removing a single-owner canonical file (brute-force / output-discipline / env-reader).
51	- Pruning during an active engagement that may still cite the content.

People who liked this also viewed

6 picks

Type

skill-prune

423stars74forks

View on GitHub

How to install skill-prune?

transilienceai/communitytools/skill-prune

$npx -y skills add transilienceai/communitytools --skill skill-prune

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Skill Prune
2
3	Inverse of `/skill-update`. Removes content rather than adding it. Run during quarterly maintenance, after engagements, or when `scripts/skill_linter.py` reports orphans / duplicates.
4
5	## When to invoke
6
7	- Quarterly cadence.
8	- After `scripts/skill_linter.py --check-orphans` reports orphan reference files.
9	- After a SKILL.md or reference file grows past its cap and needs trimming.
10	- After a de-specialization sweep, to drop content tied to retired challenges.
11
12	## Prune criteria (the four signals — same shape as /skill-update, inverted)
13
14	A reference / scenario / line is a prune candidate when it satisfies any of:
15
16	1. Orphan — not linked from any SKILL.md or other reference file in the last 60 days.
17	2. Referenced only by failed engagements — appeared in `attack-chain.md` of runs that ended `status=BLOCKED`, never in a successful chain.
18	3. Contradicted by newer content — a later scenario / pattern supersedes it; the older entry no longer reflects current technique.
19	4. Redundant with newer content — same technique covered more clearly elsewhere.
20
21	Removing content fails any of these → keep it.
22
23	## Safety rules
24
25	- Never prune a file with `<!-- KEEP: <reason> -->` annotation.
26	- Never prune content cited in a still-open engagement's `OUTPUT_DIR/attack-chain.md`.
27	- Never prune the canonical-home file for a single-owner rule (brute-force, output-discipline, env-reader, skill-update).
28	- Bias toward keeping technique-rich content over operational lore.
29
30	## Procedure
31
32	1. Run `scripts/skill_linter.py --check-orphans` to surface orphans.
33	2. For each candidate file or block, evaluate the four signals.
34	3. Build a deletion plan — show files / lines to remove with one-line rationale per item.
35	4. Apply deletions only after the plan is approved (skill-prune does not auto-delete during invocation).
36	5. Re-run `scripts/skill_linter.py` to confirm the change broke no other links and didn't reintroduce duplicates.
37
38	## Output
39
40	Concise change report:
41
42	- Removed. File or block + one-line rationale per item.
43	- Kept (flagged for follow-up). Items that failed all four signals but are worth revisiting next quarter.
44	- No prunes. State explicitly when nothing warranted removal.
45
46	## Anti-Patterns
47
48	- Pruning content because it's old — age alone is not a signal; relevance is.
49	- Pruning a reference because its parent SKILL.md is bloated — fix the SKILL.md instead.
50	- Removing a single-owner canonical file (brute-force / output-discipline / env-reader).
51	- Pruning during an active engagement that may still cite the content.

People who liked this also viewed

6 picks

Type