bastion/agentskills-publish
Skill8 starsChanged 4 months agoMITView on GitHub
Publishes a local Bastion skill to agentskills.io by stripping the bastion/ prefix, writing publish-ready frontmatter, validating with skills-ref, and guiding the git push.
- Commits and pushes
In this repository17 other files it gives its agents.
- bastion/agentskills-installSkill
- bastion/calendarSkill
- bastion-pet-packSkill
- bastion/channel-scaffoldSkill
- bastion/crisis-modeSkill
- bastion/life-logSkill
- mcporterSkill
- bastion/mesh-connectSkill
- bastion/mobile-connectSkill
- bastion/onboardingSkill
- bastion/output-validatorSkill
- bastion/persona-engineSkill
- bastion/proactive-engineSkill
- bastion/self-improvingSkill
- bastion/skill-writerSkill
- bastion/weekly-reviewSkill
- bastion/weight-systemSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 87
After writing the publish-ready frontmatter, run:
- Distinctive: line 98
If validation fails, show the error and ask the user to fix before proceeding.
- Distinctive: line 104
After validation passes, guide the user:
- Very specific: line 120
Privacy tier: only skills with privacy_tier: cloud-ok (or no privacy_tier set) are safe to publish.
- Very specific: line 129
Skill does not exist: if skills/<name>/SKILL.md is not found, ask the user to check the name and path.
- Very specific: line 133
Non-bastion/ prefix: if the skill name does not start with bastion/ (e.g., it is already a bare slug), skip the strip step and proceed …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
bastion/agentskills-publish
Guides the user through publishing a local Bastion skill to agentskills.io — handling the
bastion/ prefix strip (Pitfall 4), required metadata fields, and validation via skills-ref.
Objective
Turn a local skills/<name>/SKILL.md into a published agentskills.io entry by transforming
the frontmatter, validating the result, and walking the user through the git push step.
Flow
| Step | Action |
| 1 | Ask the user which skill to publish: "Which skill do you want to publish? (e.g., weekly-review)" |
| 2 | Read skills/<name>/SKILL.md and confirm it exists |
| 3 | Strip bastion/ prefix from the name field (see Frontmatter Transform below) |
| 4 | Write the publish-ready frontmatter, preserving metadata.bastion_name |
| 5 | Run skills-ref validate skills/<name>/SKILL.md — must pass before proceeding |
| 6 | Show the user the git push steps to submit to agentskills.io |
| 7 | Confirm the expected public URL and inform the user of the review timeline |
Frontmatter Transform (Pitfall 4)
Existing Bastion skills use name: bastion/<slug>. The agentskills.io registry rejects names
with slashes. The publish action must strip the prefix and preserve it in metadata.
Before (local SKILL.md):
---
name: bastion/my-skill
version: "1.0.0"
description: > Does something useful.
triggers:
- /my-skill
- "my trigger phrase"
---After (publish-ready frontmatter):
---
name: my-skill
version: "1.0.0"
description: > Does something useful.
triggers:
- /my-skill
- "my trigger phrase"
metadata:
bastion_name: bastion/my-skill
version: "1.0.0"
triggers:
- /my-skill
- "my trigger phrase"
---Key rules:
namefield: stripbastion/prefix → bare slug (max 64 chars, no slashes)metadata.bastion_name: preserve the originalbastion/<slug>for reinstall mappingmetadata.version: copy from theversionfieldmetadata.triggers: copy the triggers list so the hub can index by intentdescription: max 1024 chars; truncate with...if longer
Validation Step
After writing the publish-ready frontmatter, run:
skills-ref validate skills/<name>/SKILL.mdThis checks:
namehas no slashes (immediate rejection otherwise)descriptionis within 1024 chars- Required fields are present
If validation fails, show the error and ask the user to fix before proceeding.
Git Push Steps
After validation passes, guide the user:
- Fork or clone the agentskills.io skills repository
- Copy the skill directory to
skills/<name>/ - Commit:
git commit -m "add: <name> skill" - Open a pull request — the agentskills.io team reviews within ~48 h
- Once merged, the skill is available at:
https://agentskills.io/skills/<name>
Security Notes
Before publishing, check:
- No secrets in SKILL.md: API keys, tokens, passwords, and personal data must not appear in the file. SKILL.md is public after merge — treat it as public documentation.
- Privacy tier: only skills with
privacy_tier: cloud-ok(or no privacy_tier set) are safe to publish. Skills that referenceprivacy_tier: local-onlybeliefs must not be published — they may leak information about the user's local data handling. - No hardcoded URLs or internal hostnames: replace with placeholders (e.g.,
https://your-bastion.example.com).
Edge Cases
- Skill does not exist: if
skills/<name>/SKILL.mdis not found, ask the user to check the name and path. - Name already on agentskills.io: warn the user that the name exists; suggest renaming with
-v2suffix or contacting the hub maintainers to claim the existing entry. - Description too long: truncate to 1024 chars and warn the user to review the description for coherence.
- Non-bastion/ prefix: if the skill name does not start with
bastion/(e.g., it is already a bare slug), skip the strip step and proceed directly to validation.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

