@forwardimpact/outpost 3.0.3

package/templates/.claude/skills/req-forget/SKILL.md

@@ -0,0 +1,154 @@
+---
+name: req-forget
+description: >
+  Process GDPR Article 17 data erasure requests. Finds and removes all personal
+  data related to a named individual from the knowledge base, cached data, and
+  agent state files. Use when the user receives a right-to-be-forgotten
+  request, asks to delete all data about a person, or needs to comply with a
+  data erasure obligation.
+compatibility: Requires macOS filesystem access
+---
+
+# Right to Be Forgotten
+
+Process data erasure requests under GDPR Article 17. Given a person's name,
+systematically find and remove all personal data from the knowledge base, cached
+synced data, and agent state files. Produces an **erasure report** documenting
+what was found, deleted, and redacted — the compliance audit trail.
+
+## Trigger
+
+- A formal GDPR erasure request arrives.
+- The user asks to delete all data about a specific person.
+- A candidate withdraws and requests data deletion.
+- The user asks to "forget" someone.
+
+## Prerequisites
+
+- The person's full name (and any known aliases or email addresses).
+- User confirmation before any deletion.
+
+## Inputs
+
+- **Name** (required) — full name of the data subject.
+- **Aliases** (optional) — alternative names, maiden names, nicknames.
+- **Emails** (optional) — improves search coverage.
+- **Scope** — `all` (default) or `recruitment-only`.
+
+## Outputs
+
+- `knowledge/Erasure/{Name}--{YYYY-MM-DD}.md` — erasure report.
+- Deleted files and redacted references across the knowledge base.
+
+<do_confirm_checklist goal="Verify erasure is complete and the audit trail is
+sound">
+
+- [ ] User confirmed intent before any deletion.
+- [ ] All discovery recipes ran; inventory covers knowledge, cache, state,
+      drafts.
+- [ ] All dedicated files and directories deleted.
+- [ ] All mentions and backlinks redacted from other notes.
+- [ ] Cached email threads, attachments, and calendar entries handled.
+- [ ] Agent state and `graph_processed` cleaned.
+- [ ] Erasure report saved; contains **no** personal data beyond the name and
+      actions taken.
+- [ ] Final `rg` search shows only the erasure report as a match.
+
+</do_confirm_checklist>
+
+## Procedure
+
+### 0. Confirm intent
+
+State to the user:
+
+> **Data erasure request for: {Name}**
+>
+> This will permanently delete all personal data related to {Name} from:
+>
+> - Knowledge base notes (People, Candidates, Organizations mentions)
+> - Cached email threads and attachments
+> - Agent state and triage files
+>
+> This action cannot be undone. Proceed?
+
+**Wait for explicit confirmation before continuing.**
+
+### 1. Discovery
+
+Run every recipe in [references/locations.md](references/locations.md) and
+compile a complete inventory of every file and reference found.
+
+### 2. Classify
+
+For each match, pick an action from the table in
+[references/classify.md](references/classify.md). Apply its redaction rules
+where redaction is the action.
+
+### 3. Execute deletions
+
+Process most-specific to most-general.
+
+**3a. Dedicated files and directories:**
+
+```bash
+rm -rf "knowledge/Candidates/{Name}/"
+rm -f "knowledge/People/{Name}.md"
+find ~/.cache/fit/outpost/apple_mail/attachments/ -iname "*{Name}*" -delete
+```
+
+**3b. Redact mentions in other notes:** read each file, remove the specific
+lines/bullets/sections per the rules in
+[references/classify.md](references/classify.md#redaction-rules), remove broken
+`[[backlinks]]` to deleted notes, write the file back.
+
+**3c. Email threads:** delete sole-subject threads; redact paragraphs only in
+multi-person threads.
+
+**3d. Agent state files:**
+
+```bash
+for f in ~/.cache/fit/outpost/state/*_triage.md; do
+  if rg -q "{Name}" "$f" 2>/dev/null; then
+    rg -v "{Name}" "$f" > "$f.tmp" && mv "$f.tmp" "$f"
+  fi
+done
+```
+
+**3e. Processing state:** drop deleted paths from `graph_processed`:
+
+```bash
+rg -v "{deleted_path}" ~/.cache/fit/outpost/state/graph_processed \
+  > ~/.cache/fit/outpost/state/graph_processed.tmp \
+  && mv ~/.cache/fit/outpost/state/graph_processed.tmp \
+       ~/.cache/fit/outpost/state/graph_processed
+```
+
+### 4. Write the erasure report
+
+Save to `knowledge/Erasure/{Name}--{YYYY-MM-DD}.md` using the template in
+[references/report-template.md](references/report-template.md). Record **only**
+what was deleted — never CV content, skills, or assessments.
+
+### 5. Verify
+
+```bash
+rg "{Name}" knowledge/ ~/.cache/fit/outpost/ drafts/
+```
+
+The only match should be the erasure report. If other matches remain, process
+them and update the report.
+
+## Scope variants
+
+**`recruitment-only`** limits erasure to:
+
+- `knowledge/Candidates/{Name}/`
+- `knowledge/Candidates/Insights.md` mentions
+- Recruitment threads (known agency domains)
+- `recruiter_triage.md`
+
+Leaves `knowledge/People/{Name}.md` and the wider graph intact — the person may
+be a colleague or non-recruitment contact.
+
+**`all`** (default): full erasure across knowledge base, cache, and state.

package/templates/.claude/skills/req-forget/references/classify.md

@@ -0,0 +1,23 @@
+# Reference Classification
+
+Reference for `req-forget` Step 2. For every match in the inventory, pick the
+action below.
+
+| Reference Type                     | Action                                      | Example                                         |
+| ---------------------------------- | ------------------------------------------- | ----------------------------------------------- |
+| **Dedicated note** (sole subject)  | Delete entire file                          | `knowledge/People/{Name}.md`                    |
+| **Dedicated directory**            | Delete entire directory                     | `knowledge/Candidates/{Name}/`                  |
+| **Mention in another note**        | Redact: remove lines referencing the person | Backlink in `knowledge/Organizations/Agency.md` |
+| **Email thread** (sole subject)    | Delete file                                 | `~/.cache/fit/outpost/apple_mail/thread.md`     |
+| **Email thread** (multiple people) | Redact: remove paragraphs about the person  | Thread discussing multiple candidates           |
+| **Attachment** (their CV, etc.)    | Delete file                                 | `attachments/{thread}/CV.pdf`                   |
+| **Triage/state file**              | Redact: remove lines mentioning them        | `recruiter_triage.md`                           |
+| **Insights file**                  | Redact: remove bullets mentioning them      | `knowledge/Candidates/Insights.md`              |
+
+## Redaction rules
+
+- Remove entire bullet points that mention the person by name.
+- Remove table rows containing the person's name.
+- Remove `## Connected to` entries linking to deleted notes.
+- If a section becomes empty after redaction, remove its header too.
+- Do **not** remove surrounding context that doesn't identify the person.

package/templates/.claude/skills/req-forget/references/locations.md

@@ -0,0 +1,59 @@
+# Discovery Search Recipes
+
+Reference for `req-forget` Step 1. Run every recipe; record every match into the
+inventory.
+
+## Knowledge base — direct notes
+
+```bash
+ls -d "knowledge/Candidates/{Name}/" 2>/dev/null
+ls "knowledge/People/{Name}.md" 2>/dev/null
+
+# Common name variations
+ls "knowledge/People/{First} {Last}.md" 2>/dev/null
+ls "knowledge/People/{Last}, {First}.md" 2>/dev/null
+```
+
+## Knowledge base — backlinks and mentions
+
+```bash
+rg -l "{Name}" knowledge/
+rg -l "{First name} {Last name}" knowledge/
+rg -l "\[\[.*{Name}.*\]\]" knowledge/
+rg -l "{email}" knowledge/
+```
+
+## Cached email threads
+
+```bash
+rg -l "{Name}" ~/.cache/fit/outpost/apple_mail/ 2>/dev/null
+rg -l "{email}" ~/.cache/fit/outpost/apple_mail/ 2>/dev/null
+find ~/.cache/fit/outpost/apple_mail/attachments/ -iname "*{Name}*" 2>/dev/null
+```
+
+## Cached calendar events
+
+```bash
+rg -l "{Name}" ~/.cache/fit/outpost/apple_calendar/ 2>/dev/null
+rg -l "{email}" ~/.cache/fit/outpost/apple_calendar/ 2>/dev/null
+```
+
+## Agent state files
+
+```bash
+rg -l "{Name}" ~/.cache/fit/outpost/state/ 2>/dev/null
+```
+
+## Drafts
+
+```bash
+rg -l "{Name}" drafts/ 2>/dev/null
+```
+
+## Final verification
+
+```bash
+rg "{Name}" knowledge/ ~/.cache/fit/outpost/ drafts/
+```
+
+Expected: only the erasure report matches.

package/templates/.claude/skills/req-forget/references/report-template.md

@@ -0,0 +1,67 @@
+# Erasure Report Template
+
+Audit trail for `req-forget` Step 4. Save to
+`knowledge/Erasure/{Name}--{YYYY-MM-DD}.md`.
+
+**The report itself must not contain personal data** beyond the subject's name
+and the actions taken. Do not copy CV content, skills, or assessments into the
+report — record only what was deleted, not what it contained.
+
+````markdown
+# Data Erasure Report — {Full Name}
+
+**Date:** {YYYY-MM-DD HH:MM}
+**Requested by:** {user or "GDPR Article 17 request"}
+**Scope:** {all / recruitment-only}
+
+## Data Subject
+- **Name:** {Full Name}
+- **Known aliases:** {aliases or "none"}
+- **Known emails:** {emails or "none"}
+
+## Actions Taken
+
+### Deleted Files
+- `knowledge/Candidates/{Name}/brief.md`
+- `knowledge/Candidates/{Name}/CV.pdf`
+- `knowledge/Candidates/{Name}/screening.md`
+- `knowledge/People/{Name}.md`
+- {list all deleted files}
+
+### Redacted References
+- `knowledge/Organizations/{Agency}.md` — removed backlink
+- `knowledge/Candidates/Insights.md` — removed {N} bullet(s)
+- {list all redacted files and what was removed}
+
+### Cached Data Removed
+- `~/.cache/fit/outpost/apple_mail/{thread}.md` — deleted (sole subject)
+- `~/.cache/fit/outpost/apple_mail/{thread2}.md` — redacted (multi-person)
+- {list all cache actions}
+
+### State Files Cleaned
+- `~/.cache/fit/outpost/state/recruiter_triage.md` — redacted
+- {list all state file actions}
+
+## Requires Manual Action
+
+The following are outside this tool's reach:
+
+- **Apple Mail** — original emails remain in the user's mailbox. Search
+  Mail.app for "{Name}" and delete threads manually.
+- **Apple Calendar** — events remain in Calendar.app.
+- **Recruitment agencies** — notify {Agency} of the deletion and request
+  they do the same.
+- **Interview notes** — check physical notebooks and external apps.
+- **Shared documents** — check Google Drive, SharePoint, etc.
+
+## Verification
+
+```bash
+rg "{Name}" knowledge/ ~/.cache/fit/outpost/
+````
+
+Expected: no matches except this erasure report.
+
+```
+
+```

package/templates/.claude/skills/req-scan/SKILL.md

@@ -0,0 +1,145 @@
+---
+name: req-scan
+description: >
+  Scan publicly available sources for candidates who indicate they are open for
+  hire. Uses WebFetch to read public APIs (HN Algolia, GitHub, dev.to).
+  Writes prospect notes to knowledge/Prospects/. Maintains
+  cursor/dedup state in ~/.cache/fit/outpost/head-hunter/. Use when the
+  head-hunter agent is woken or when the user asks to scan for open candidates.
+---
+
+# Scan Open Candidates
+
+Fetch and filter publicly available candidate posts from platforms where people
+**explicitly indicate** they are open for hire. This skill handles fetching,
+filtering, deduplication, benchmarking, prospect-note writing, and memory
+updates.
+
+## Trigger
+
+- The head-hunter agent is woken by the scheduler.
+- The user asks to scan for open candidates or prospects.
+
+## Prerequisites
+
+- `WebFetch` tool available (Claude Code built-in — no curl/wget).
+- `fit-pathway` CLI available (`bunx fit-pathway`).
+- Memory directory `~/.cache/fit/outpost/head-hunter/`.
+
+## Inputs
+
+- `~/.cache/fit/outpost/head-hunter/cursor.tsv` — source rotation.
+- `~/.cache/fit/outpost/head-hunter/seen.tsv` — deduplication.
+- Standard data via `bunx fit-pathway skill --list` and `bunx fit-pathway job`.
+
+## Outputs
+
+- `knowledge/Prospects/{Name}.md`.
+- Updated `cursor.tsv`, `seen.tsv`, `prospects.tsv`, and `log.md`.
+- `~/.cache/fit/outpost/state/head_hunter_triage.md`.
+
+<do_confirm_checklist goal="Verify the wake produced a clean, ethical scan">
+
+- [ ] Selected the least-recently-checked, non-suspended source.
+- [ ] Fetched via `WebFetch` (never curl/wget).
+- [ ] Applied all 5 filters in order; passed candidates have ≥ 2
+      standard-relevant skills.
+- [ ] Deduplicated against `seen.tsv` before processing.
+- [ ] Benchmarked each prospect against a real `bunx fit-pathway job`.
+- [ ] Prospect notes follow the template (no fabricated names).
+- [ ] All state mutations went through `node scripts/state.mjs` — cursor, seen,
+      prospects, failures, log.
+- [ ] Triage report written to the state directory; alternative queries (if any)
+      logged.
+- [ ] Failures recorded; sources with ≥ 3 consecutive failures suspended.
+
+</do_confirm_checklist>
+
+## Procedure
+
+### 1. Pick a source
+
+```bash
+node scripts/state.mjs cursor list
+```
+
+Pick the least-recently-checked source. Skip suspended sources (`failure get` ≥
+3). If all are suspended, report it in the triage and exit. Source URLs and
+parse fields: [references/sources.md](references/sources.md).
+
+### 2. Fetch and scan
+
+Use `WebFetch` per [references/sources.md](references/sources.md). On failure,
+follow [references/filters.md](references/filters.md#failure-handling).
+
+### 3. Filter
+
+Apply the 5-filter pipeline in
+[references/filters.md](references/filters.md#filter-pipeline) — signal, dedup,
+geographic, skill alignment, experience level.
+
+### 3b. Fallback — zero new prospects
+
+If filtering eliminates every candidate, try up to **3 alternatives** per wake
+from [references/fallbacks.md](references/fallbacks.md). Log every alternative
+in `log.md`.
+
+### 4. Benchmark and write prospect notes
+
+For each candidate that passes filters, benchmark:
+
+```bash
+bunx fit-pathway job {discipline} {estimated_level} --track={best_track}
+```
+
+Classify match strength:
+
+- **strong** — multiple core skills match, level aligns, location works, plus
+  non-traditional signals for forward-deployed.
+- **moderate** — some overlap, level roughly right, minor gaps.
+- **weak** — few matching signals, significant gaps.
+
+Write notes for **strong** and **moderate** matches only, using the template in
+[references/template.md](references/template.md).
+
+```bash
+mkdir -p "knowledge/Prospects"
+```
+
+### 5. Update state
+
+All state changes go through `node scripts/state.mjs` — full command reference:
+[references/state.md](references/state.md). Each wake:
+
+1. Update the cursor (`cursor set`).
+2. Reset failure count on success or increment on failure.
+3. Append every processed post ID to `seen` (`seen batch` for many).
+4. Add new prospects (`prospect add`).
+5. Append the wake summary (`log-wake`).
+
+### 6. Write the triage report
+
+Save to `~/.cache/fit/outpost/state/head_hunter_triage.md`:
+
+```markdown
+# Head Hunter Triage — {YYYY-MM-DD HH:MM}
+
+## Last Scan
+Source: {source_id} ({description})
+Posts scanned: {N}
+New prospects: {N}
+Skipped: {N} (dedup: {N}, location: {N}, skill fit: {N})
+Alternative queries tried: {N} ({list, or "none needed"})
+
+## Pipeline Summary
+Total prospects: {N} (strong: {N}, moderate: {N})
+Sources checked today: {list}
+Oldest unchecked source: {source_id} (last: {date})
+Suspended sources: {list with failure counts, or "none"}
+
+## Recent Prospects
+- **{Name}** — {match_strength}, {estimated_level} {track}, {location}
+
+## Retention
+{List prospects older than 90 days not acted on, if any.}
+```

package/templates/.claude/skills/req-scan/references/fallbacks.md

@@ -0,0 +1,76 @@
+# Fallback Strategies
+
+Reference for `req-scan` when a primary source query yields zero new prospects
+after filtering. Try at most **3 alternatives per wake** to stay within rate
+limits.
+
+## Strategy 1 — alternative search terms
+
+### HN
+
+- Check the **previous** month's "Who Wants to Be Hired?" thread.
+- Search for `"Who is hiring"` threads (candidates sometimes post in the wrong
+  thread; comments may link to candidate profiles).
+- Try
+  `https://hn.algolia.com/api/v1/search?query=%22freelancer+available%22&tags=comment`.
+
+### GitHub
+
+Search by skill + availability:
+
+```
+WebFetch URL: https://api.github.com/search/users?q=%22data+engineering%22+%22open+to+work%22&per_page=30&sort=joined&order=desc
+WebFetch URL: https://api.github.com/search/users?q=%22full+stack%22+%22available+for+hire%22&per_page=30&sort=joined&order=desc
+WebFetch URL: https://api.github.com/search/users?q=%22devops%22+%22looking+for%22&per_page=30&sort=joined&order=desc
+```
+
+Search repos with README signals:
+
+```
+WebFetch URL: https://api.github.com/search/repositories?q=%22hire+me%22+in:readme&sort=updated&order=desc&per_page=10
+```
+
+Try alternate location terms: Greece, Athens, Warsaw, Bucharest, Sofia,
+Manchester, Edinburgh.
+
+### dev.to
+
+Try broader tags: `jobsearch`, `career`, `remotework`, `job`, `hiring`. Or pull
+from a tag and filter by title/description:
+
+```
+WebFetch URL: https://dev.to/api/articles?tag=career&per_page=25
+```
+
+## Strategy 2 — relax filters
+
+If geography eliminated everyone, re-scan without the location filter. Mark
+unmatched-location prospects as "location unconfirmed" in the prospect note.
+
+If skill alignment filtered everyone out: lower the bar from 2 standard skills
+to 1; consider transferable skills (strong Python ⇒ likely data integration
+capability); accept adjacent indicators ("machine learning" implies data
+skills).
+
+## Strategy 3 — cross-reference
+
+When a source yields very few results, enrich what you do find:
+
+- A GitHub profile that links to a blog/portfolio → WebFetch it for more detail
+  before deciding skill fit.
+- An HN post that mentions a GitHub username → fetch their GitHub profile for
+  richer signal.
+
+## Logging alternatives
+
+Log every alternative in `log.md`:
+
+```markdown
+## 2026-03-05 14:00
+
+Source: github_open_to_work
+Primary query: "open to work" location:UK — 30 results, 0 new after dedup
+Alternative 1: "data engineering" "open to work" — 12 results, 1 new prospect
+Alternative 2: "full stack" "available for hire" — 8 results, 0 new
+Stopped after 2 alternatives (1 prospect found)
+```

package/templates/.claude/skills/req-scan/references/filters.md

@@ -0,0 +1,83 @@
+# Filter Pipeline & Failure Handling
+
+Reference for `req-scan` Steps 3 (filter) and 5 (failure handling).
+
+## Filter pipeline
+
+Apply in order. Stop at the first that excludes the candidate.
+
+### 1. Open-for-hire signal
+
+- HN "Who Wants to Be Hired?" → **auto-pass** (thread is opt-in).
+- GitHub → must have `hireable: true` **or** bio contains an open-to-work
+  phrase.
+- dev.to → must be tagged `opentowork` or `lookingforwork`.
+
+### 2. Deduplication
+
+```bash
+node scripts/state.mjs seen check {source} {post_id}
+```
+
+If exit 0 (seen), skip. Otherwise continue.
+
+### 3. Geographic fit
+
+Prefer or accept candidates in or open to:
+
+- **US East Coast** (NYC, Boston, DC, Philadelphia, …).
+- **UK** (London, Manchester, Edinburgh, …).
+- **EU** — especially Greece, Poland, Romania, Bulgaria.
+- **Remote / Anywhere / Global**.
+
+Skip candidates explicitly locked to incompatible regions ("San Francisco only",
+"APAC only"). When location is ambiguous or unstated, include the candidate —
+let the user decide.
+
+### 4. Skill alignment
+
+`bunx fit-pathway skill --list` for the standard inventory. Strong signals:
+
+**Forward Deployed:** multiple industries / domains, customer-facing projects,
+data integration / analytics / visualization, full-stack with business context,
+non-traditional path (law, policy, academia → tech), AI/ML tool proficiency
+(Claude, GPT, Cursor, "vibe coding").
+
+**Platform:** infrastructure / cloud / DevOps, architecture and system design,
+API design and shared services, performance / scalability / reliability.
+
+**Minimum bar:** at least 2 standard-relevant skills must be identifiable from
+the post. Skip candidates with purely non-technical profiles.
+
+### 5. Experience level
+
+| Signal                                      | Likely level |
+| ------------------------------------------- | ------------ |
+| "junior", "entry-level", 0–2 years          | J040         |
+| "mid-level", 3–5 years                      | J060         |
+| "senior", 5–8 years, "lead"                 | J070         |
+| "staff", "principal", 8+ years, "architect" | J090+        |
+
+When uncertain, default to J060 with low confidence.
+
+## Failure handling
+
+When a WebFetch fails (HTTP 4xx, 5xx, timeout, empty response, redirect to a
+block page):
+
+1. Increment the failure count:
+   `node scripts/state.mjs failure increment {source}`.
+2. **Do not retry** the same source in this wake cycle.
+3. Suspend after 3 consecutive failures — skip during source selection. Retry
+   suspended sources once every 7 days to detect recovery.
+4. Log the failure in `log.md` with HTTP status and error.
+5. Reset on success: `node scripts/state.mjs failure reset {source}`.
+
+### Common failure patterns
+
+| Symptom                  | Likely cause                                           |
+| ------------------------ | ------------------------------------------------------ |
+| 503 with HTML redirect   | Corporate proxy blocking the domain (won't recover)    |
+| 403 Forbidden            | Auth required or automated requests blocked            |
+| 429 Too Many Requests    | Rate limited — will recover; don't suspend permanently |
+| Empty response / timeout | Transient — likely recovers next wake                  |

package/templates/.claude/skills/req-scan/references/sources.md

@@ -0,0 +1,76 @@
+# Sources
+
+Reference for `req-scan` Step 2 (fetch & scan). One source per wake cycle.
+
+## 1. Hacker News "Who Wants to Be Hired?"
+
+Monthly thread, posted on the 1st.
+
+```
+WebFetch URL: https://hn.algolia.com/api/v1/search?query=%22Who+wants+to+be+hired%22&tags=ask_hn&hitsPerPage=5
+```
+
+The first hit whose title matches "Who wants to be hired?" with `created_at` in
+the current or previous month is the target thread.
+
+```
+WebFetch URL: https://hn.algolia.com/api/v1/items/{objectID}
+```
+
+`children[]` contains top-level comments — each is one candidate. Parse the
+`text` field for:
+
+- **Location** ("Location: New York", "NYC / Remote").
+- **Remote** ("Remote: Yes", "Open to remote").
+- **Skills** — tech stack, language/standard mentions.
+- **Experience** — years, role titles, past companies.
+- **Contact** — email (often obfuscated: "name [at] domain [dot] com").
+- **Resume / CV** — links to personal sites, GitHub, LinkedIn.
+
+**Cursor:** thread `objectID` plus the ID of the last processed child comment.
+**Rate:** no strict limit; one fetch per wake cycle.
+
+## 2. GitHub Open to Work
+
+Search by location (rotate one query per wake):
+
+```
+WebFetch URL: https://api.github.com/search/users?q=%22open+to+work%22+location:UK&per_page=30&sort=joined&order=desc
+WebFetch URL: https://api.github.com/search/users?q=%22open+to+work%22+location:Europe&per_page=30&sort=joined&order=desc
+WebFetch URL: https://api.github.com/search/users?q=%22looking+for+work%22+location:remote&per_page=30&sort=joined&order=desc
+```
+
+Alternate bio phrases to rotate across wakes: `"available for hire"`,
+`"seeking opportunities"`, `"seeking new role"`, `"open to new opportunities"`,
+`"currently exploring"`, `"freelance" "available"`, `"between roles"`,
+`"on the market"`, `"open to opportunities"`.
+
+Fetch each promising candidate's full profile:
+
+```
+WebFetch URL: https://api.github.com/users/{login}
+```
+
+Profile fields: `name`, `bio`, `location`, `hireable`, `blog`, `company` (null =
+likely available), `public_repos`, `created_at`.
+
+**Cursor:** location query last used + page number. Rotate UK → Europe → Remote
+→ repeat. **Rate:** 10 requests/min unauthenticated. Fetch at most 5 full
+profiles per wake (1 search + 5 profile fetches = 6 requests).
+
+## 3. dev.to
+
+```
+WebFetch URL: https://dev.to/api/articles?tag=opentowork&per_page=25
+WebFetch URL: https://dev.to/api/articles?tag=lookingforwork&per_page=25
+```
+
+Fields: `title`, `description`, `user.name`, `user.username`, `url`, `tag_list`,
+`published_at`. Skip articles older than 90 days.
+
+**Cursor:** `id` of the most recent article processed. **Rate:** 30 req per 30
+s. One fetch per wake.
+
+## Use WebFetch only
+
+**Never use `curl` or `wget`.** WebFetch is the supported transport.