Breakdown tasks skill

[trmartin4]

📔 Objective

Adds the next in a sequence of changes to decompose and enhance the existing writing-tech-breakdowns skill, for use with our new tech-breakdowns repo.

The plan for skill decomposition is roughly defined below, and this is the 3rd skill in that plan:

The plan for skill decomposition is roughly defined below, and these are the first 2 skills in that plan:

• Creating a new breakdown Breakdown initialization and spec definition skills #141
• Defining the Specification (the WHAT and WHY) Breakdown initialization and spec definition skills #141
• Defining the Plan (the HOW) Breakdown plan skill #143
• Defining the Tasks (this PR)
• Completing the breakdown (archiving to avoid ambiguous future references)

[github-actions]

✅ Plugin Validation Summary

All checks passed. Two plugins were validated for this PR: bitwarden-delivery-tools (bumped to 2.0.0) and bitwarden-tech-lead (bumped to 2.3.2). No critical, major, or blocking issues were found. A small number of minor warnings are noted below as optional polish.

This PR's substantive change is a coordinated breaking refactor: two breakdown skills were removed from bitwarden-delivery-tools and a new decomposing-into-tasks skill was added, with bitwarden-tech-lead updated to drop stale references. The version bumps and changelogs correctly reflect this.

Changed components

Component	Change	Plugin
skills/coordinating-cross-team-breakdown/	Deleted	bitwarden-delivery-tools
skills/writing-tech-breakdowns/SKILL.md	Deleted	bitwarden-delivery-tools
skills/decomposing-into-tasks/	Added	bitwarden-delivery-tools
skills/developing-breakdown-plan/SKILL.md	Modified	bitwarden-delivery-tools
skills/developing-breakdown-spec/SKILL.md	Modified	bitwarden-delivery-tools
skills/navigating-the-initiative-funnel/SKILL.md	Modified	bitwarden-delivery-tools
agents/AGENT.md, plugin.json, CHANGELOG.md	Modified	bitwarden-tech-lead

---

1. Plugin Structure Validation (plugin-validator) — ✅ PASS

• plugin.json manifests — both valid: kebab-case names, valid semver (2.0.0 / 2.3.2), well-formed required fields, parse cleanly.
• Version/name consistency — marketplace.json, plugin.json, CHANGELOG.md, and AGENT.md all agree for both plugins.
• CHANGELOG.md — Keep a Changelog + SemVer format; bumps are correctly sized (delivery-tools MAJOR for breaking skill removal; tech-lead PATCH for doc/reference-only realignment).
• Agent frontmatter (bitwarden-tech-lead/agents/AGENT.md) — name valid (19 chars, lowercase-hyphen), 4 well-formed <example> blocks, model: opus, color: cyan, substantial system prompt; referenced skills (architecting-solutions, contributing-to-technical-strategy) exist.
• Cross-plugin reference integrity — AGENT.md references the current breakdown skills (starting-breakdown, developing-breakdown-spec, developing-breakdown-plan, decomposing-into-tasks), all of which exist. No stale references to the removed skills remain in any live file (only in CHANGELOG history, which is correct).
• README / file organization — present and accurate; README skill tables match the 10 skills on disk.

2. Skill Quality Review (skill-reviewer) — ✅ PASS

The two SKILL.md files reported as "missing" by the reviewer are intentional deletions in this PR (confirmed via git diff: coordinating-cross-team-breakdown/SKILL.md and writing-tech-breakdowns/SKILL.md are D status). These are not errors — they are documented under the delivery-tools 2.0.0 MAJOR bump.

The four existing reviewed skills all pass:

• decomposing-into-tasks — valid frontmatter, ~1,600 words, references/process-flow.dot exists. ✅
• developing-breakdown-plan — valid frontmatter, ~1,700 words, references/process-flow.dot + references/worked-example.md exist. ✅
• developing-breakdown-spec — valid frontmatter, ~960 words, explicit read-only MCP tool scoping, references/process-flow.dot exists. ✅
• navigating-the-initiative-funnel — valid frontmatter, ~1,580 words, all Skill(...) cross-references resolve. ✅

All cross-skill Skill(...) references resolve; all referenced references/ files exist; word counts are within/near target; progressive disclosure is correctly applied.

3. Security Validation (claude-config-validator) — ✅ PASS

• No committed secrets — no API keys, tokens, passwords, or private keys. The single grep hit (AGENT.md:26, "org-scoped tokens are issued") is benign domain prose.
• No hardcoded credentials.
• Tool scoping is least-privilege — skill allowed-tools lists are narrow; MCP tools are read-only Atlassian operations (get_*, search_*) with no create/update/delete. Agent uses Read, Write, Glob, Grep, Skill.
• No dangerous command auto-approvals — no rm -rf, curl | sh, eval, sudo, or --dangerously* patterns (the eval matches are the word in eval-set filenames, not code execution).

---

Minor warnings (optional — non-blocking)

Severity	File:Line	Note
Minor	navigating-the-initiative-funnel/SKILL.md:3	Description leads descriptively; consider adding 2–3 verbatim quoted trigger phrases to match sibling skills' trigger quality.
Minor	navigating-the-initiative-funnel/SKILL.md:7	Missing top-level # H1 / ## Overview heading used by the other breakdown skills (consistency only).
Minor	developing-breakdown-spec/SKILL.md:32	Carries a ToolSearch/deferred-tool harness instruction for TaskCreate that the other skills omit; reconcile (preferably remove the harness detail everywhere).
Minor	decomposing-into-tasks/SKILL.md:60	Grammar: "the tasks satisfies" → "the tasks satisfy".
Minor	decomposing-into-tasks/SKILL.md:108–158	Large inline code-example block is a candidate for examples/; not required (word count in range).
Minor	developing-breakdown-plan / navigating-the-initiative-funnel	Cross-plugin Skill(architecting-solutions) / Skill(bitwarden-security-context) calls create install-time dependencies; consider noting them in the plugin README so standalone installs don't hit unresolved skill calls.

---

Verdict: ✅ Release-ready

Both plugins are valid, version-consistent, credential-free, and the breaking change is correctly coordinated across both with appropriately sized version bumps and accurate changelogs. The flagged items are minor, optional polish — none block this PR.