run-maintenance-task
Use this skill to execute a task from the Android Agentic Maintenance Backlog. Invoke when the user provides an Asana task URL from that backlog and asks to run, start, execute, or work on it. Also invoke when the user says things like "run the next maintenance task", "work on this backlog item", or "pick up a maintenance task". Requires the Asana MCP to be configured.
93
91%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Advisory
Suggest reviewing before use
Quality
Discovery
89%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This description excels at trigger term coverage and completeness, with explicit 'Invoke when' clauses and natural user phrases. Its main weakness is that the 'what' portion is somewhat vague — it says 'execute a task' without specifying what concrete actions that entails (e.g., reading requirements, making code changes, updating task status). The description also uses second-person 'Use this skill' framing at the start, though the rest is appropriately impersonal.
Suggestions
Add specific concrete actions describing what 'executing a task' involves, e.g., 'Reads task requirements from Asana, implements code changes, runs tests, and updates task status.'
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | It names the domain (Android Agentic Maintenance Backlog, Asana tasks) and describes the general action (execute a task), but doesn't list specific concrete actions like what executing a task actually entails (e.g., reading task details, updating status, running scripts). | 2 / 3 |
Completeness | Clearly answers both 'what' (execute a task from the Android Agentic Maintenance Backlog) and 'when' (explicit trigger guidance with 'Invoke when...' and 'Also invoke when...' clauses with example phrases). Also mentions a prerequisite (Asana MCP). | 3 / 3 |
Trigger Term Quality | Includes strong natural trigger terms: 'Asana task URL', 'run', 'start', 'execute', 'work on', 'run the next maintenance task', 'work on this backlog item', 'pick up a maintenance task'. These are phrases users would naturally say. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive — scoped to a specific backlog ('Android Agentic Maintenance Backlog'), a specific tool (Asana MCP), and specific trigger patterns (Asana task URLs). Unlikely to conflict with other skills. | 3 / 3 |
Total | 11 / 12 Passed |
Implementation
92%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a high-quality workflow skill that provides clear, actionable instructions for executing Android maintenance tasks. Its greatest strengths are the explicit validation checkpoints, concrete commands with fallback alternatives, and well-structured error recovery flow. The only minor weakness is that it's somewhat long for a single file and references external resources (CLAUDE.md, ddg-asana skill) without bundle support, though this is a minor concern given the workflow nature of the skill.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is lean and efficient throughout. Every section serves a clear purpose, there's no explanation of concepts Claude already knows (git, Asana, PRs), and the instructions are direct commands rather than tutorials. The fallback patterns (gh vs REST API) are necessary operational detail, not padding. | 3 / 3 |
Actionability | Provides specific, executable commands throughout: exact git worktree commands, gradle commands, gh API calls with method and endpoint, PR title format, and exact PR body content. Fallback commands are also concrete with full API paths. The PR template filling instructions are specific enough to be copy-paste actionable. | 3 / 3 |
Workflow Clarity | The workflow is clearly sequenced across well-defined phases (Before you start → Implement → Verify → Open PR → If stuck). Validation checkpoints are explicit: check task state before starting, run spotless and tests before PR, fix failures before proceeding, and the 'If stuck' section provides a clear error recovery path with cleanup steps. The 'Do not open a PR with known failures' constraint is an explicit gate. | 3 / 3 |
Progressive Disclosure | The content is well-organized with clear section headers, but it's a moderately long monolithic file. References to external files (CLAUDE.md, .github/PULL_REQUEST_TEMPLATE.md, ddg-asana skill) are mentioned but there are no bundle files to support them. Some content like the PR template filling instructions or the full fallback API patterns could potentially be split out, though the single-file approach is reasonable for a workflow skill. | 2 / 3 |
Total | 11 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- duckduckgo/Android
- Commit
adccd8d
- Reviewed
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.