Conversation
Co-authored-by: openhands <openhands@all-hands.dev>
all-hands-bot
left a comment
There was a problem hiding this comment.
🟡 Acceptable - Solves a real problem (agents being asked to "modify yourself"), but SKILL.md is too dense for progressive disclosure. Main issue: section 3 turns into a uv technical manual when it should stay a decision guide.
Co-authored-by: openhands <openhands@all-hands.dev>
enyst
left a comment
There was a problem hiding this comment.
Addressed the review feedback:
- Trimmed SKILL.md section 3 to stay high-level and moved details to references/INSTALLATION_MODES.md.
- Clarified MCP config locations: CLI uses ~/.openhands/mcp.json (or $OPENHANDS_PERSISTENCE_DIR/mcp.json); SDK skills/plugins use per-skill/per-plugin .mcp.json.
- Updated hooks template to parse structured JSON with jq and return {decision: allow|deny}.
- Reworded the opening clarification guidance to avoid unnecessary “one at a time” ceremony.
Pushed in 52633f7.
Co-authored-by: openhands <openhands@all-hands.dev>
|
Rename update:
|
all-hands-bot
left a comment
There was a problem hiding this comment.
🟡 Acceptable - Solves a real problem and the structure is sound after previous revisions. Main issue: one overly-generic trigger that will cause false positives.
skills/adapt-yourself/SKILL.md
Outdated
| - .agents/skills | ||
| - hooks.json | ||
| - plugin.json | ||
| - uv tool install |
There was a problem hiding this comment.
🟠 Important: The trigger "uv tool install" is too generic and will cause false positives. This skill is about self-modification/customization, but this trigger would fire whenever someone mentions uv tool install for any purpose (e.g., "How do I uv tool install ruff?" or "Install pytest with uv").
Why this matters: Trigger pollution degrades the progressive disclosure model. The skill will get loaded unnecessarily, adding context bloat to unrelated conversations.
Fix: Remove the "uv tool install" line from the triggers list. The other triggers ("modify yourself", "customize openhands", "self-modify", etc.) are specific enough to catch actual self-modification requests without false positives.
If you genuinely need to catch uv-related self-modification, rely on the combination of existing triggers + the natural context of the conversation, not a generic command string.
Co-authored-by: openhands <openhands@all-hands.dev>
|
Pulled your latest tweaks (
Pushed in 3c3536c. |
Co-authored-by: openhands <openhands@all-hands.dev>
|
Updated section “Add tools via MCP” to be more restrictive:
Pushed in 5063dcb. |
Co-authored-by: openhands <openhands@all-hands.dev>
|
Pulled latest (
Pushed in 682ba1e. |
Co-authored-by: openhands <openhands@all-hands.dev>
|
Did a Changes:
Pushed in 4e5e8ca. |
Adds a new skill (
adapt-yourself) that teaches agents/users how to make persistent behavior changes by choosing the right extensibility mechanism:.agents/skills)AGENTS.md).openhands/hooks.json)~/.openhands/mcp.json; SDK skills/plugins: per-skill.mcp.json)Also includes references covering uv +
uv tool installworkflows and version pinning, plus templates.Changes:
skills/adapt-yourself/(SKILL.md, README, references).plugin/marketplace.jsonValidation:
python -m json.tool .plugin/marketplace.json