Compress Skills
Compress one or more SKILL.md files so they consume less context on load while preserving 100% of their value — same triggers, capabilities, guardrails, thresholds, and output. Use when the user says 'compress this skill', 'optimize skills for context', 'make skills use less context', 'slim down the skills', 'reduce skill token usage', 'shrink SKILL.md', or wants a skill or whole skill suite trim…
Install
$ agentstack add skill-therapys-dotagents-compress-skills ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Links to your public security report.[](https://agentstack.voostack.com/security/report/skill-therapys-dotagents-compress-skills) Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Compress Skills
Shrink the tokens a skill costs when loaded, with zero loss of value: every trigger, step, guardrail, tool name, threshold, and output spec survives. This is compression, not redesign — never add features or change behavior.
How skills spend context (optimize the right layer)
- Frontmatter
description— loads into every session; drives auto-invocation. Keep all trigger phrases/keywords/scenarios. You may tighten wording, never drop a trigger. Editing it is high-risk; touch it last and least. - SKILL.md body — loads on every invocation. Primary target.
- Reference files (
references/*.md,REFERENCE.md, recipes) — load only when opened. Move occasionally-needed detail here to shrink the mandatory body.
Process
- Scope. List the target skills.
wc -ceach file for a baseline. Skip files gitignored as private unless the user names them. Skip already-minimal files ( ~30% to confirm no capability was dropped.
Guardrails
- Triggers are sacred. A dropped trigger phrase silently breaks discovery. Diff descriptions to prove they're intact.
- Don't touch protected literals — code, HogQL/SQL, CSS/color/radius/shader values, exact commands/flags/paths stay byte-for-byte.
- Leave already-tight files alone. Trimming a 500 B skill risks value for no real saving.
- Respect
.gitignore. Private, local-only skills stay untouched unless the user names them. - Report honestly. Give per-file numbers, flag anything uncertain, and don't inflate the reduction.
Report format
Per file: : B → B (−XX%), a TOTAL line, 3–6 bullets on what was cut/moved and why value survives, and a RISK line. A shareable summary (e.g. display.dev) is optional — only when asked.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: therapys
- Source: therapys/dotagents
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.