| name | accomplishments-review |
| description | Use when preparing for a job search, updating a resume, prepping for interviews, or doing a performance review. Triggered by "review accomplishments", "/accomplishments-review", "prepare for interviews", "update my resume", "performance review prep". |
Accomplishments Review
Synthesize the accomplishments log into job-search-ready outputs.
Step 0 — Locate files and calibrate
Read memory/me.md for the workspace path.
- Log directory:
{{WORKSPACE_PATH}}/_accomplishments/ — year-based files (2025.md, 2026.md, etc.)
- Weekly summaries:
{{WORKSPACE_PATH}}/daily-notes/ — files named YYYY-WW.md
Ask (or infer from context):
- Output type: resume bullets, interview stories, performance review, or all three?
- Time range: all time, last N months, or a specific date range?
- Focus area: all projects, or a specific theme?
- Role level: if not set, ask — this determines the importance bar.
Importance bar by level:
| Level | What clears the bar | What to drop |
|---|
| Junior / IC1-2 | Delivered on a commitment, owned a task end-to-end, proactively flagged a problem | Work done entirely under direction with no independent contribution |
| Mid / IC3-4 | Took full ownership of a project, drove a metric, navigated ambiguity, contributed across teams | Tasks completed but not owned; work that required no independent judgment |
| Senior / IC5 | Made a call that changed direction, shipped something with measurable business impact, unblocked cross-team work | Routine execution, tickets completed without broader impact |
| Staff / Principal | Set or shifted strategy, resolved contested issues with lasting impact, drove alignment at exec level, shaped how the team thinks about a problem area | Docs written, meetings attended, tickets moved, routine reviews |
| Executive | Org outcomes, hiring decisions, culture shifts, decisions that set direction for multiple teams | Individual project execution, anything below the org or strategic level |
When in doubt: "Could a more junior person have done this, and would it be forgettable in 6 months?" If yes, drop it.
Step 1 — Close open loops
Scan two sources for open loops:
Prior review files: Glob _accomplishments/review-*.md. For each, look for entries that were in-progress, unresolved, or flagged.
Prior year files: If the time range starts mid-year or later, also read the year file(s) immediately before the range start. Scan for entries where:
- A project was described as in-progress or building toward something
- A metric was flagged as "TBD"
- Alignment or a decision was described as pending
- A proposal or initiative had not yet landed
These are cross-year open loops — work that started in one year file and may have resolved in another.
Present all open loops as a short list and ask for closure on any the user can provide. Add confirmed outcomes to the current output under a "Closed loops" section before proceeding.
Step 2 — Gather raw material
Read two sources in parallel:
Source A: Year files — Glob _accomplishments/????.md and read only the files whose year falls within the requested time range. For each entry extract: date, project/area, action verb, impact, metrics, draft bullet.
Source B: Weekly summaries — Glob daily-notes/**/YYYY-WW.md within the time range. For each file, extract the ### What I drove this week section only. These are already filtered for personal agency — use them as corroborating signal and as a source for entries that may not have been formally logged.
If a weekly summary item has no corresponding log entry, flag it as a logging gap and include it in the output using the weekly summary text as the draft.
Step 3 — Filter by importance
Apply the calibration from Step 0. For each raw entry, assign:
- Include — clears the bar
- Weak — marginal; include only if the log is thin, otherwise drop
- Drop — routine execution
For Staff/Principal: be aggressive. 10 strong entries beat 25 mediocre ones.
Step 4 — Group by theme
Cluster included entries into 4–6 themes based on what's actually in the log. For a PM, suggested groupings (adjust to fit):
- Product strategy and definition — specs, PRDs, prioritization calls, roadmap decisions
- Cross-functional alignment — driving agreement, unblocking stuck decisions
- Customer and business impact — features shipped, adoption driven, revenue-connected work
- Data and analytics — measurement frameworks, analytics features, data-driven decisions
- Process and tooling — workflows, documentation, team infrastructure
- Platform / AI — platform strategy, data model decisions, AI-native features
Step 5 — Generate outputs
Resume bullets
For each theme, produce 2–4 polished bullets:
- Lead with a strong action verb
- Include the most specific metric available
- One line, ~100 characters
- No internal jargon — readable by an external recruiter
**[Theme]**
• [Bullet]
• [Bullet]
Flag any metric placeholder: ⚠️ needs metric.
Interview stories (STAR format)
Select 3–5 entries with the clearest impact, personal agency, and narrative arc. For each:
**[Story title]**
**Situation:** [2–3 sentences: what was happening, why it mattered]
**Task:** [What you were specifically responsible for]
**Action:** [3–5 sentences: your specific choices and actions]
**Result:** [Lead with the metric or most concrete outcome, then qualitative impact]
**Good for:** [Interview question types, e.g. "Tell me about a time you drove alignment"]
Performance review summary (if requested)
One-page narrative:
- What I shipped — bullet list of delivered work
- How I drove impact — 2–3 paragraphs on business outcomes
- How I worked with others — cross-functional contributions
- What I'm building toward — optional forward-looking paragraph
Step 6 — Flag gaps
After generating outputs, note:
- Entries with "metric TBD" worth following up on
- Themes with fewer than 2 entries (underlogged or genuinely thin?)
- Time gaps of 2+ weeks in the log (quiet period or logging missed?)
- Weekly summary items with no log entry (candidates for back-logging)
Step 7 — Save
Offer to save the full output — closed loops, resume bullets, interview stories, performance review (if requested), and flags — to _accomplishments/review-YYYY-MM-DD.md.
