Attribution

This plugin is shaped by work shared across the agent-tooling community. The sections below credit the projects, practitioners, standards, and official documentation behind specific skills, workflows, and conventions.

Copied prose, scripts, templates, and substantial assets must be permitted by a verified upstream license and ship with its complete required notice, not only an attribution in this README. Adapted workflows should record their source in the skill's references/sources.yaml, rewrite the workflow in this plugin's style, and avoid direct ports of long monolithic skill bodies. See THIRD_PARTY_NOTICES.md for retained third-party material and its skill-local notices.

From addyosmani/agent-skills

• trackbed:code:work-from-plan, trackbed:code:plan-to-pr, and trackbed:linear:issue-implement context preparation, plus the shared forward-progress reference's context recovery: Adapted from context-engineering.
• trackbed:docs:feature-spec: Adapted from spec-driven-development.
• trackbed:docs:adr: Adapted from documentation-and-adrs.
• trackbed:code:deprecate: Adapted from deprecation-and-migration.
• trackbed:code:api-design contract-first rules, retry safety with idempotency keys, error shape and status mapping, naming conventions, pagination, REST patterns, and TypeScript interface patterns: Adapted from api-and-interface-design.
• trackbed:code:encode-constraint bypass guards, shrink-only baselines, entry-point red runs, and check-script exit statuses: Informed by constraint-driven-development.
• trackbed:debug:triage-to-issue Prove-It regression-test conventions: Adapted from test-driven-development.
• trackbed:debug:investigate and trackbed:debug:triage-to-issue shared root-cause procedure (reproduce, isolate, trace, and bisect guidance): Adapted from debugging-and-error-recovery.
• trackbed:browse security boundaries, JavaScript constraints, content boundary markers, and Addy marker conventions: adapted from browser-testing-with-devtools.
• trackbed:qa network triage ladder, clean-console standard, accessibility ladder, and Addy marker conventions: adapted from browser-testing-with-devtools.
• trackbed:git:commit-message, trackbed:pr:generate-description, trackbed:git:recreate-commits, trackbed:pr:rebase: Addy output markers, the Change Summary triplet (CHANGES MADE / THINGS I DIDN'T TOUCH / POTENTIAL CONCERNS) in trackbed:git:recreate-commits, 6-item pre-commit checklist, and "dev branches are costs" framing adapted from git-workflow-and-versioning. Addy's per-commit Conventional Commits rule is explicitly rejected.

From EveryInc/compound-engineering-plugin

• trackbed:architecture-strategist, trackbed:design-iterator, and trackbed:performance-oracle: Adapted from the plugin's review agents.
• trackbed:product:strategy: Adapted from skill ce-strategy.
• trackbed:code:optimize: Adapted from ce-optimize, reviewed at commit 6f9ab03a031c054a8046659926251fb6c149269f.
• trackbed:session:automate-repeats: Adapted from skill ce-sessions, including its safe session discovery and extraction scripts reviewed at commit 6f9ab03a031c054a8046659926251fb6c149269f.
• trackbed:setup: Informed by skill ce-setup, first reviewed at commit 6f9ab03a031c054a8046659926251fb6c149269f; its workflow and checker are original.
• trackbed:docs:solution-note (including --refresh): Adapted from skills ce-compound and ce-compound-refresh, reviewed at commit 6f9ab03a031c054a8046659926251fb6c149269f.
• trackbed:docs:review: Adapted from ce-doc-review.
• trackbed:code:work-from-plan: Adapted from ce-work and ce-plan as a thin routing adapter, not a full autonomous execution pipeline.
• trackbed:linear:issue-implement showing the issue title before the first write, tool discovery before reporting Linear unavailable, the stop after two failed fixes, and pending user action for acceptance criteria outside the repository: Informed by ce-work, reviewed at commit 7dc50943bbf959dbe6a93a175e41e5fff8395e00.
• trackbed:pr:triage-risk handling of deleted files and quiet failures: Informed by ce-code-review pull requests #1719 and #1888, reviewed at commit 18c2be5e3d79024f20bce936b8cb05272621760a.
• trackbed:code:api-design Design It Twice framing rules and final-pick verification: Informed by ce-bakeoff, reviewed at commit 6e7655a1ccf359e9e82cdcbdf38fa37a4ec680e1.
• Codex converter: Inspired by the plugin's converter approach.
• External-source adaptation policy, copied-script attribution guardrails, and artifact-lifecycle prompts: Informed by the repository at commit 6f9ab03a031c054a8046659926251fb6c149269f, including representative skills ce-compound, ce-compound-refresh, ce-plan, ce-code-review, and ce-optimize.
• Shared dev-server detection scripts and browser-facing auto URL detection contract: Adapted from ce-polish, reviewed at commit 6f9ab03a031c054a8046659926251fb6c149269f.

From google/elephant-goldfish-model

Reviewed at commit 403d8881d62022490637d740657b33f01e321ea8, the Codex integration guide at 4ac6b173c15704339bcd654f22a16b4c30120385, and the readiness standard and style reviewer at 6efc127defc561e3a672b3d155b8216255a10e25. These are ideas rewritten in this plugin's own words; no upstream text is retained.

• trackbed:cold-reader and its launch contract: Informed by its Claude Code and Codex integration guides.
• The marker technique in scripts/probe-cold-reader-isolation.js and the probe's run record: Informed by its isolation probe and harness test record.
• The comprehension read in trackbed:docs:review and trackbed:siw:spec-audit: Informed by its comprehension reviewer.
• The opt-in false-compromise read (--trade-offs) in trackbed:docs:review: Informed by its inventive catalyst reviewer, reviewed at commit 6a8001ea31ccf72430ba0603a467ec6df9f69dd7; trackbed writes its own pattern list.
• The current-state claims check in trackbed:codebase-pattern-reviewer and the path:line citation check in scripts/check-doc-citations.py: Informed by its codebase details reviewer.
• The trackbed:siw:spec-audit checks for file boundaries, decisions left to the implementer, defensive additions, and caveat remedies: Informed by its readiness standard, review design, and style reviewer.
• The Caveats and Claims test in the shared plain-language rubric that trackbed:text:clarify and trackbed:pr:generate-description apply: Informed by its style reviewer.
• The failure-signal, rollback, and running-cost checks in trackbed:siw:spec-audit and trackbed:docs:review: Informed by the cost, migration, and observability lenses in egm-brainstorm.
• Rejected approaches in trackbed:docs:feature-spec: Informed by egm-draft.
• Where trackbed:docs:adr may take deciders from: Informed by the rejection-attribution change to egm-draft in #249, reviewed at commit de27902a154c4a92dd52aac50082d53873e6ac91.
• trackbed:workflow:suggest-next: Informed by the draft EGM Ask design from #218, reviewed at commit 394aa38db6ceb641d4e37adc552e0a56535aa093.
• The contest and escalate rules in trackbed:siw:resolve-audit and trackbed:pr:resolve-review: Informed by egm-resolve.

Other sources

• trackbed:verify:run evidence-before-claims rule and red-green check: Inspired by obra/superpowers' verification-before-completion skill.
• trackbed:debug:triage-to-issue: The hands-off bug-to-issue workflow is adapted from Matt Pocock's triage-issue, while its vertical RED-GREEN cycle guidance is adapted from Pocock's TDD skill.
• Module-depth guidance in trackbed:code:agent-readiness, trackbed:code:refactor-opportunities, trackbed:code:api-design, trackbed:pr:code-review with trackbed:architecture-strategist, trackbed:pr:triage-risk, and the spec, issue, and plan templates: Informed by AI Hero's How To Make Codebases AI Agents Love. The shared architecture glossary adapts the vocabulary of Matt Pocock's codebase-design skill.
• trackbed:code:refactor-opportunities hot-spot seeding, deepening patterns, planned-change mode (--for), and one-item hand-off: Informed by Matt Pocock's improve-codebase-architecture skill and AI Hero's guide to it.
• Name-collision guidance in the shared implementation context checklist that trackbed:linear:issue-implement, trackbed:code:work-from-plan, and trackbed:code:plan-to-pr apply, and the search-uniqueness checks in trackbed:code:agent-readiness: Informed by Modem's How AI coding agents read your code.
• trackbed:code:audit-practice: The eighteen mistakes it checks for are adapted from Matt Pocock's post on agentic coding mistakes, with terms read through AI Hero's AI Coding Dictionary.
• trackbed:code:encode-constraint: Inspired by @poteto's X post on treating the codebase as agent memory and turning each observed agent mistake into a constraint.
• trackbed:product:prd: The PRD written after prototyping as a decision record, with its behavior contract, golden-set evals, rollout gates, kill switch, and change log, is adapted from Aakash Gupta's post on an AI-era PRD template.
• trackbed:docs:update-agents-md: Inspired by getsentry/skills.
• trackbed:git:commit-message: From getsentry/skills.
• trackbed:text:clarify and the shared plain-language rubric that trackbed:pr:generate-description also applies: Inspired by fofr's GOV.UK style skill and official GOV.UK guidance on identifying user needs and using clear language, adapted into a locale-neutral reader-task workflow.
• trackbed:text:clarify --humanize: Its AI-writing pattern catalog adapts Wikipedia: Signs of AI writing (maintained by WikiProject AI Cleanup) under CC BY-SA 4.0 and retains adapted text from blader/humanizer's SKILL.md at commit 3e1da8bfce43ecbe43a56f3cb00b62fbdf1cbd56 under the MIT License. Both notices ship with the skill.
• trackbed:visual:explain, including its --onboarding mode: Adapted from nicobailon/visual-explainer.
• trackbed:visual:check-slop: Vendors and adapts Gesso Build's MIT-licensed anti-slop detector and 73-rule registry at commit 1c3908b7efb56ec24624436dea5c32a371eb487c. Complete notices for Gesso and bundled parser dependencies ship with the skill.
• Skills authoring patterns: Inspired by mgechev/skills-best-practices.
• trackbed:session:automate-repeats effectiveness evidence, counterfactual improvement gate, and explicit skill-use extraction: Inspired by Warp's skill-doctor workflow, rubrics, guidance, and collector at commit 737129f.
• trackbed:pr:github-review-reply: GitHub review comment listing, review-summary reads, top-level PR comment reads/posts, reply posting, review thread mapping, and thread resolution operations are grounded in official GitHub REST and GraphQL API documentation.
• trackbed:pr:github-review: review-requested PR discovery, PR-context reads, pull/<N>/head fetch, existing-conversation mapping (REST review comments, GraphQL review threads, issue comments, prior reviews), and the optional review-submission/reply appendix are grounded in the GitHub CLI manual and official GitHub REST/GraphQL/search documentation.