The public community skills library for Trinity agents — the default skills-library source a Trinity instance syncs out of the box.
Each skill is a directory under skills/<name>/ with a SKILL.md carrying the library's frontmatter contract — the SKILL.md format is harness-portable, so the catalog uses a vendor-neutral layout declared in catalog.yaml (skills root, category enum, credential-provider registry) rather than any harness's dotfolder. Trinity instances sync this repo pinned to a release tag, surface it in the Skills tab, and inject assigned skills into agents. Every skill also works in a plain Claude Code session — copy its directory into your agent's .claude/skills/.
Status: seeding. The v1 catalog (22 skills across 5 categories) is being promoted in from its authoring sources. The first release tag (v0.1.0) will be cut when the seed lands — until a v* tag exists, instances cannot pin this repo as a source.
| Skill |
Version |
Origin |
Description |
add-git-sync |
1.3 |
mirrored · abilities |
Add git-as-state hooks to an agent — auto-commits on Stop, rebases on SessionStart, snapshots on PreCompact |
add-memory |
1.1 |
mirrored · abilities |
Add a memory system to an agent — file awareness, knowledge graph, structured state, or multi-session tracking |
add-pipeline |
1.8 |
mirrored · abilities |
Scaffold a Trinity-compatible long-running pipeline inside any agent — creates projects/<slug>/{project.md, pipeline.yaml, instances/} |
adjust-playbook |
1.13 |
mirrored · abilities |
Modify an existing playbook based on conversation context or explicit instructions |
agent-fleet-analysis |
2.5 |
mirrored · abilities |
Scan one or more directories of agents in ANY paradigm — Claude Code, n8n workflow exports, framework apps (LangChain/CrewAI/AutoGen), or fr |
agent-fleet-migrate |
1.2 |
mirrored · abilities |
Build a verified Claude Code fleet from an agent-fleet-analysis work order — non-destructive migration of mixed fleets (Claude Code, n8n exp |
create-playbook |
2.19 |
mirrored · abilities |
Create a new skill or playbook |
validate-pipeline |
1.3 |
mirrored · abilities |
Lint a pipeline.yaml — schema check, DAG acyclicity, referenced-skill existence, precondition kind registration |
| Skill |
Version |
Origin |
Description |
document-extractor |
1.0 |
library |
Extracts key information from documents (PDFs, images, text files) in a folder and creates structured markdown summaries with the same filen |
epub-chapter-extractor |
1.0 |
library |
Extract all chapters from an EPUB file into separate markdown files |
file-indexer |
1.0 |
library |
Generate a comprehensive file system index with directory tree, file sizes, and modification dates |
| Skill |
Version |
Origin |
Description |
autoplan |
1.0 |
mirrored · abilities |
Analyze a skill issue before implementing — reads the affected SKILL.md, identifies what changes are needed and any risks |
backlog |
1.1 |
mirrored · abilities |
Show current GitHub Issues backlog — what's in progress, what's next, priorities |
claim |
1.0 |
mirrored · abilities |
Claim the next issue from the backlog — picks highest priority todo, moves it to in-progress |
close |
1.0 |
mirrored · abilities |
Close the current issue without a git commit — adds summary comment and marks done |
commit |
1.1 |
mirrored · abilities |
Commit changed skill files and close the in-progress issue — writes a traceability commit message referencing the issue number |
groom |
1.0 |
mirrored · abilities |
Groom the backlog — tag untagged issues with skill:* labels, verify priorities, surface stale in-progress work |
project-init |
1.1 |
mirrored · abilities |
Create or adopt a long-term managed project per PROJECT_STANDARD.md — GitHub epic issue with idempotent label creation and a project_files/< |
project-intake |
1.2 |
mirrored · abilities |
Headless intake primitive — routes actionable items from any source (meetings, email, Slack, issue trackers) into the GitHub Issues registry |
project-reconcile |
1.1 |
mirrored · abilities |
Sync projection adapters against the GitHub Issues registry per PROJECT_STANDARD.md |
project-steward |
1.3 |
mirrored · abilities |
Autonomous sweep of all managed projects per PROJECT_STANDARD.md |
project-task |
1.3 |
mirrored · abilities |
Create a task issue in the uniform format per PROJECT_STANDARD.md — the ONLY sanctioned task-creation path |
roadmap |
1.0 |
mirrored · abilities |
Strategic view of the agent backlog — open issues grouped by skill, showing which areas have the most work |
sprint |
1.0 |
mirrored · abilities |
Human-supervised development cycle — orchestrates roadmap → claim → autoplan → implement → commit for one skill issue |
work-loop |
1.3 |
mirrored · abilities |
Autonomous work loop — pick one backlog issue, execute it, close it, exit |
| Skill |
Version |
Origin |
Description |
gemini-critic |
1.3 |
library |
Ask Gemini's most powerful model for an independent second opinion - critique code with fresh eyes, evaluate an article, challenge a plan |
repo-velocity |
1.5 |
library |
Measure the development speed / velocity / activity of any GitHub repository using objective metrics from the GitHub API — commits, lines ad |
| Skill |
Version |
Origin |
Description |
animated-explainer |
1.26 |
library |
Build an animated, data-driven explainer film for a real system — a canvas film that renders live from the product's own data and cannot dri |
create-explanatory-image |
1.1 |
library |
Generate explanatory diagrams and infographics that visually communicate concepts |
explain-visually |
1.1 |
library |
Explain a complex topic back to you with visuals - the right diagram for each facet (flowchart, timeline, sequence, schema/ER, state machine |
microsite |
1.5 |
library |
Build a brand-locked, animated single-page microsite (locked to YOUR palette - the shipped tokens are a swappable default) - a self-containe |
nano-banana-image-generator |
1.0 |
library |
Generate images using Google's Nano Banana 2 (Gemini 3.1 Flash Image Preview) |
one-pager |
1.2 |
library |
Generate a brand-locked, information-dense single-page PDF that lets a reader understand an entire topic - a company, a product, a project, |
presentation |
1.2 |
library |
Generate a brand-locked multi-slide presentation deck (16:9, 1920x1080) and assemble into a PDF |
| Skill |
Version |
Origin |
Description |
self-diagnostic |
1.0 |
library |
Run self-diagnostics on the agent to verify skills, commands, agents, and dependencies are working |
skill-builder |
1.0 |
library |
Expert guide for creating Claude Code skills |
update-dashboard |
1.1 |
library |
Refresh an agent's business metrics in one pass — compute every metric declared in template.yaml metrics:, refresh dashboard.yaml, and recor |
workspace-discipline |
1.0 |
library |
Enforce workspace organization rules when creating files, writing documents, or generating assets |
See CONTRIBUTING.md — the conventions (layout, naming, categories, versioning, credentials), the trust model (merged ≠ shipped: releases are tags), and the review bar. tools/validate.py is the executable form of the contract; CI runs it on every push and PR.
Apache-2.0