oh-my-agent-commandcode 0.0.0-stage → 20.38.4

package/skills/oma-image/resources/vendor-matrix.md

@@ -0,0 +1,90 @@
+# Vendor Matrix
+
+## Reference Image Support (`--reference` / `-r`)
+
+| Vendor | Reference input | Transport | Notes |
+|--------|-----------------|-----------|-------|
+| `codex` | PASS | `codex exec -i <path>` (repeatable) | Local file path; 5MB-per-file cap enforced by Codex CLI |
+| `antigravity` | PASS | Per-run temp dir staged via `agy --add-dir`; paths referenced inline in prompt | Up to 10 refs (skill cap); copied so agy sandboxes don't need access to the originals' parent dir |
+| `pollinations` | N/A | (none) | Requires URL hosting; rejected with exit 4 when explicitly selected, silently dropped from the run set under `--vendor auto`. Planned for PR #2. |
+
+All paths are validated in `reference-guard.ts` (magic-byte MIME check + size + count + duplicate rejection) before dispatch. The magic-byte-detected MIME is threaded through `GenerateInput.referenceImages` and used verbatim at the vendor API boundary; file extension is never trusted for MIME type.
+
+## Codex
+
+| Field | Value |
+|-------|-------|
+| Binary | `codex` (npm: `@openai/codex`) |
+| Auth | OAuth via `codex login` |
+| Health check | `codex login status` output contains "Logged in" |
+| Model | `gpt-image-2` |
+| Transport | `codex exec --json "<instruction>"` (internal bridge invokes `image_gen` tool; the first JSONL event carries the thread id) |
+| Image location | `$CODEX_HOME/generated_images/<thread_id>/*.png` (`CODEX_HOME` defaults to `~/.codex`) → copied to `outDir`. Without a thread id, images new since the run started are taken from the whole folder. |
+| Sizes | Any `WxH` passing `size-guard.ts` (each edge ∈ [16, 3840], multiples of 16, aspect 1:3..3:1) or `auto`. Codex CLI clamps to its own gpt-image-2 limits internally. |
+| Qualities | `low`, `medium`, `high`, `auto` |
+
+Codex requires `--skip-git-repo-check` for invocation inside a git worktree; this is inherited from the upstream `codex-image` skill and is a known dependency of the Codex CLI image path.
+
+## Antigravity
+
+The Antigravity CLI (`agy`) is an agentic CLI that runs against the user's Gemini Code Assist subscription (no separate API key, no per-image charge). Since 1.2.16, it routes native image requests through the built-in `image-generator` subagent; older sessions expose `generate_image` directly. OMA supports both routes and waits for generated image bytes to be saved or copied to the requested target paths. Image model selection remains internal: the prompt has no model hint, the manifest records `"model": "agy-internal"`, and the output filename is `antigravity-<runShortid>.<ext>` with no model segment.
+
+| Field | Value |
+|-------|-------|
+| Binary | `agy` (Antigravity CLI) |
+| Auth | Sign-in via Gemini Code Assist account during `agy install` |
+| Health check | `agy --version` exits 0 — **install check only**. Sign-in state is not verified; a signed-out agy passes health and fails at generate time. |
+| Model selection | Opaque — chosen by agy's internal agent loop. Not exposed via flags, not recorded as a vendor-side promise. |
+| Transport | `agy -p --dangerously-skip-permissions --add-dir <outDir> --print-timeout <s> "<instruction>"` (spawn `cwd` is forced to `<outDir>` to prevent agy from inheriting a stale workspace context) |
+| Output bytes | Saved directly by agy to absolute paths embedded in the prompt; detected via PNG/JPEG/WebP/GIF magic bytes and renamed accordingly |
+| Sizes | Any `WxH` allowed by `size-guard.ts` (each edge ∈ [16, 3840], multiples of 16, aspect 1:3..3:1) or `auto`. The dimension is passed as an advisory hint in the prompt — agy/Gemini may pick its own internal aspect. |
+
+### Why `agy` instead of the Gemini CLI / direct API
+
+- `gemini -p` runs the full agent loop and does not emit raw `inlineData` bytes on stdout (as of Gemini CLI 0.38) — it tries to invoke image-generation tools itself, often recursing back into `oma-image`.
+- The direct `generativelanguage.googleapis.com` API requires `GEMINI_API_KEY` plus billing on AI Studio. Image models are not in the free tier.
+- `agy -p` already wraps Gemini Code Assist credentials in the user's Antigravity session and provides native image generation. It writes raw bytes to disk for us, so we don't have to capture them from stdout.
+
+### Output format gotcha
+
+Gemini image surfaces currently return JPEG bytes regardless of the requested filename extension. The provider writes to a `.img` placeholder, sniffs the actual format from the first 12 bytes, then renames to `.png` / `.jpg` / `.webp` / `.gif`. The result's `mime` field reflects the sniffed format, not the user's requested extension.
+
+## Pollinations
+
+| Field | Value |
+|-------|-------|
+| Binary | (none — direct HTTP) |
+| Auth | `POLLINATIONS_API_KEY` env var (free key at https://enter.pollinations.ai) |
+| Health check | `POLLINATIONS_API_KEY` is set |
+| Models | Free (no pollen credits): `flux`, `zimage`. Credit-gated (positive pollen balance required): `qwen-image`, `wan-image`, `gptimage`, `gptimage-large`, `gpt-image-2`, `klein`, `kontext`. Select via `--model`; default `flux`. |
+| Transport | `POST https://gen.pollinations.ai/v1/images/generations` (OpenAI-compatible), `response_format: b64_json` |
+| Sizes | Any `WxH` (non-`WxH` or `auto` falls back to `1024x1024`) |
+| Output | PNG or JPEG, sniffed from magic bytes |
+
+A credit-gated model without pollen balance fails as `invalid-input` (exit 4) with a hint to switch to `flux`/`zimage`.
+
+## Strategy Attempt Record
+
+Manifest field `strategy_attempts` is always an array of objects:
+
+```
+{ "strategy": "codex-exec-oauth" | "agy-print" | "pollinations-http",
+  "status": "ok" | "skipped" | "failed",
+  "reason"?: string,
+  "duration_ms"?: number }
+```
+
+The last successful strategy also appears on the run as `strategy`.
+
+## Error Classification
+
+| Error kind | Retry policy | Exit code when solo |
+|------------|-------------|---------------------|
+| `not-installed` | fail (the vendor health check catches this first) | 5 |
+| `auth-required` | fail; printed hint tells user how to authenticate | 5 |
+| `invalid-input` | fail; surfaces validation problems from the provider | 4 |
+| `safety-refused` | short-circuit (no fallback) | 2 |
+| `rate-limit` | record attempt as failed | 1 (if no vendor succeeded) |
+| `timeout` | record attempt as failed | 6 |
+| `network` | `retryable=true` → record; `false` → record | 1 |
+| `other` | record | 1 |

package/skills/oma-market/SKILL.md

@@ -0,0 +1,142 @@
+---
+name: oma-market
+description: Research customer pain points, trends, and competitors through the
+  OMA market engine. Use for market discovery or voice-of-customer analysis.
+---
+
+# Market Research Agent - Community Signal Intelligence
+
+## Scheduling
+
+### Goal
+Run the upstream `last30days` research engine (always the latest release, managed by oma) for community-signal research, then frame the result for the user's intent (pain / trend / competitor / discovery) with strategic frameworks and save one brief under `.agents/results/market/`.
+
+### Intent signature
+- User asks about pain points, user complaints, or voice-of-customer signals for a product or category.
+- User asks what is trending, growing, or declining in a space this week or month.
+- User asks how one product compares to another in community sentiment or positioning.
+- User asks for discovery or exploratory market research on a topic, a person, a company, or a ticker.
+
+### When to use
+- Extracting real user pain points from community posts (Reddit with real upvotes and top comments, HN, X, Bluesky, GitHub Issues)
+- Detecting trends in a category over a window (`--days 7|30|90|180`)
+- Competitor sentiment analysis and SWOT / Porter's 5F positioning
+- Open-ended discovery research (`--discover`), person mode, hiring signals (`--hiring-signals`), follow-up drills (`--drill`)
+
+### When NOT to use
+- General web research without market framing -> use oma-search directly
+- Academic literature -> use oma-scholar
+- Live dashboards or scheduled monitoring -> `oma schedule <action>` wrapping this skill
+
+### Expected inputs
+- Topic string; optional `--intent pain|trend|competitor|discovery` (else classified per `resources/intent-rules.md`)
+- Optional window (`--days`), `--vs <entity>` (competitor), `--frameworks auto|none|swot,5f,pestel`
+- Any native last30days flag (see `oma market run --help`) — passed through verbatim
+
+### Expected outputs
+- Single markdown brief at `.agents/results/market/{topic-slug}-{YYYYMMDD}.md`
+- First line: the engine's badge (`🌐 last30days v{VERSION} · synced {date}`); body per the upstream OUTPUT CONTRACT; framework sections appended per intent; engine footer preserved
+- Raw engine artifacts under `market.save_dir` (default `.agents/results/market/raw/`)
+
+```yaml
+outputs:
+  - name: market-brief
+    description: Single LAW-compliant markdown brief with framework sections
+    artifact: ".agents/results/market/*.md"
+    required: true
+```
+
+### Dependencies
+- `oma market resolve` / `oma market run` — engine location, Python 3.12+ resolution, `--save-dir` default
+- The upstream `SKILL.md` at the resolved engine root (`skillMd` in `oma market resolve --json`) — the authoritative research contract
+- `resources/intent-rules.md`, `resources/frameworks/`, `resources/output-laws.md`, `resources/execution-protocol.md`
+
+### Control-flow features
+- `oma market detect-trap` gate before anything else (exit 2 = REFUSE, exit 4 = invalid)
+- Engine is the skills-CLI install of `last30days`: `oma market resolve` reads installed skill dirs (a `market.path` / `LAST30DAYS_HOME` pin wins); `oma market update` installs it when missing and updates it otherwise
+- Sources needing keys/cookies auto-skip inside the engine; keyless sources (Reddit, HN, GitHub, Polymarket, arXiv, Techmeme, Digg, web) always run
+- Framework auto-toggle by intent (pain/trend → SWOT; competitor → SWOT + Porter's 5F; discovery → SWOT + PESTEL)
+
+## Structural Flow
+
+### Entry
+1. Run `oma market detect-trap "<topic>"`. Exit 2 → surface the REFUSE reason and reframe suggestion, stop.
+2. Run `oma market resolve --json`. `ok: false` → report `reason` (missing engine → ask the user, then `oma market update`; missing Python → the install hint) and stop. Never fall back to WebSearch-only synthesis and present it as market research.
+3. Read the upstream contract at `engine.skillMd` **top to bottom**. It is long by design; do not skim. Treat `engine.root` as its `SKILL_DIR`.
+4. Classify intent per `resources/intent-rules.md`; map to engine flags and framework set.
+
+### Scenes
+1. **PREPARE**: detect-trap, resolve, read upstream SKILL.md, classify intent.
+2. **UPSTREAM STEPS**: follow the upstream SKILL.md exactly — Step 0 (first-run setup wizard, consent-driven), intent parsing, Step 0.45 (its own query-quality preflight), Step 0.5 / 0.55 (handle, subreddit, hashtag resolution when WebSearch is available), Step 0.75 (query plan). Skip only its "Runtime Preflight" Python-hunt block: `oma market run` already resolved the interpreter.
+3. **RUN**: wherever the upstream contract says `"${LAST30DAYS_PYTHON}" "${SKILL_DIR}/scripts/last30days.py" <args>`, run `oma market run <args>` with the **same arguments** (foreground, 5-minute timeout, `--emit=compact`). `--save-dir` is added automatically from `market.save_dir` unless you pass one.
+4. **SYNTHESIZE**: produce the brief exactly as the upstream OUTPUT CONTRACT dictates (badge first line, Ranked Evidence Clusters, LAWs). Then append the framework sections selected for the intent, using only clusters present in the engine output as evidence (`resources/frameworks/`).
+5. **FINALIZE**: run the self-check in `resources/output-laws.md`, write `.agents/results/market/{topic-slug}-{YYYYMMDD}.md`, preview the first 50 lines.
+
+### Transitions
+- `--vs <entity>` or "A vs B" phrasing → competitor intent → upstream COMPARISON flow (two passes + head-to-head as its contract specifies) → SWOT + Porter's 5F.
+- Person / company / ticker topics → upstream person / hiring-signals / StockTwits handling applies unchanged.
+- `engine.status: stale` → include the `note` in the report (research ran on the cached engine version).
+
+### Failure and recovery
+- detect-trap exit 2 → REFUSE; do not run the engine; `--force` only on explicit user reconfirmation.
+- `oma market resolve` not ok → stop with the reason; no engine run.
+- Engine non-zero exit → report stderr verbatim; do not synthesize from partial stdout unless the upstream contract says the emitted compact output is still valid.
+- Upstream Python-version gate / setup wizard messages → relay to the user exactly as the upstream contract instructs.
+
+### Exit
+- Success: brief written with badge, clusters, frameworks, and engine footer; path reported.
+- Partial: engine ran with skipped sources (footer lists them) — say so; never pad with invented evidence.
+
+## Logical Operations
+
+### Actions
+| Action | SSL primitive | Evidence |
+|--------|---------------|----------|
+| detect-trap preflight | `VALIDATE` | Topic arg, trap pattern rules |
+| Resolve engine + Python | `CALL_TOOL` | `oma market resolve --json` |
+| Read upstream contract | `READ` | `engine.skillMd` |
+| Classify intent | `SELECT` | `resources/intent-rules.md` |
+| Upstream pre-research steps | `INFER` | Upstream SKILL.md Steps 0–0.75 |
+| Run engine | `CALL_TOOL` | `oma market run <args>` |
+| Synthesize + frameworks | `WRITE` | Upstream OUTPUT CONTRACT, `resources/frameworks/` |
+| Self-check + write brief | `WRITE` | `resources/output-laws.md`, `.agents/results/market/` |
+
+### Tools and instruments
+- `oma market detect-trap <topic>` (preflight gate)
+- `oma market resolve [--json]` (engine + Python resolution from installed skill dirs)
+- `oma market update` (install the skill when missing, otherwise `bunx skills update last30days -g -y`)
+- `oma market run <engine args…>` (passthrough to the resolved upstream engine’s Python entry point)
+
+### Canonical command path
+```bash
+TOPIC="VS Code pain points"
+oma market detect-trap "$TOPIC"
+oma market resolve --json            # read .engine.skillMd, then follow it
+# … upstream Steps 0 / 0.45 / 0.5 / 0.55 / 0.75 …
+oma market run "$TOPIC" --plan "$QUERY_PLAN_FILE" --subreddits=vscode --emit=compact --save-suffix=v3
+```
+
+### Resource scope
+| Scope | Resource target |
+|-------|-----------------|
+| `NETWORK` | Inside the engine only (its per-source fetchers); GitHub for the managed engine refresh |
+| `LOCAL_FS` | installed skill dir such as `~/.agents/skills/last30days/` (engine), `~/.config/last30days/` (engine config, keys), `.agents/results/market/` (brief + raw) |
+| `PROCESS` | `oma market` subcommands → the resolved upstream Python entry point |
+
+### Preconditions
+- Topic passes detect-trap.
+- `oma market resolve` is ok (engine present; Python ≥ 3.12 found on PATH, via `uv`, or pinned with `market.python` / `LAST30DAYS_PYTHON`).
+
+### Effects and side effects
+- Writes the brief to `.agents/results/market/{topic-slug}-{YYYYMMDD}.md` and raw engine files to `market.save_dir`.
+- First run: the upstream setup wizard may write `~/.config/last30days/.env` with user consent.
+- If Python 3.12+ is absent, `oma market resolve` stops before the upstream setup wizard. Relay its install hint; after an authorized interpreter installation, retry resolution. Do not expect the skipped upstream Python preflight to install it.
+
+## References
+- Execution protocol: `resources/execution-protocol.md`
+- Intent routing: `resources/intent-rules.md`
+- Output contract: `resources/output-laws.md`
+- Applicable framework: `resources/frameworks/swot.md`, `resources/frameworks/porters-5f.md`, or `resources/frameworks/pestel.md` (load only the selected framework)
+- Validation: `resources/checklist.md`
+- Recovery: `resources/error-playbook.md`
+- Upstream engine instructions: read the resolved `engine.skillMd` path from `oma market resolve --json`; upstream scripts are managed engine files, not bundled skill resources.

package/skills/oma-market/resources/checklist.md

@@ -0,0 +1,12 @@
+# oma-market checklist
+
+- [ ] `oma market detect-trap "<topic>"` ran first; exit 2 surfaced, not bypassed
+- [ ] `oma market resolve --json` is `ok`; engine version and `status` noted (`stale` → mention)
+- [ ] Upstream `SKILL.md` (`engine.skillMd`) read top to bottom this run
+- [ ] Upstream Step 0 (setup wizard) honoured on first run; user consent for any `.env` / Python install
+- [ ] Upstream Steps 0.45 / 0.5 / 0.55 / 0.75 done when WebSearch is available, else `--auto-resolve`
+- [ ] Engine ran via `oma market run … --emit=compact` in the foreground
+- [ ] Intent classified; framework set matches `intent-rules.md` (or user override)
+- [ ] Brief: badge first line, engine structure untouched, frameworks cite only engine clusters
+- [ ] `output-laws.md` self-check clean
+- [ ] Written to `.agents/results/market/{slug}-{YYYYMMDD}.md`; path + skipped sources reported

package/skills/oma-market/resources/error-playbook.md

@@ -0,0 +1,12 @@
+# oma-market error playbook
+
+| Symptom | Cause | Action |
+|---|---|---|
+| `detect-trap` exit 2 | keyword-trap / demographic-shopping / too-broad topic | Show REFUSE + reframe; stop. `--force` only after explicit user reconfirmation |
+| `oma market resolve` → `ok: false`, "not installed" | last30days skill not installed | with the user's consent, `oma market update` (runs `bunx skills add mvanhorn/last30days-skill --skill last30days -g -y`), or set `market.path` / `LAST30DAYS_HOME` |
+| `resolve` → "Python 3.12+" hint | no compatible interpreter | relay hint (`brew install python@3.12` / `apt install python3.12` / `uv python install 3.12`) or set `market.python` / `LAST30DAYS_PYTHON`; stop, never WebSearch-only |
+| `oma market update` → `failed` | skills CLI / network / `bunx` missing | relay `detail`; the installed copy (if any) still runs |
+| Engine prints setup-wizard prompts | first run, no `~/.config/last30days/.env` | follow upstream Step 0 in chat; never answer consent questions on the user's behalf |
+| Engine non-zero exit | source/network/auth failure | report stderr verbatim; check `oma market run "<topic>" --diagnose` |
+| Sources listed as skipped in footer | keys / cookies absent | keep the list visible; suggest the upstream setup wizard for those sources |
+| Output has `Sources:` block, `—`, `##` in body | LAW violation during synthesis | fix per upstream LAWs and `output-laws.md` self-check before writing |

package/skills/oma-market/resources/execution-protocol.md

@@ -0,0 +1,60 @@
+# Step-by-step execution protocol for oma-market
+
+oma owns the gate, the engine lifecycle, and the framework layer. The upstream
+`last30days` SKILL.md owns everything about the research itself. Never
+improvise a research flow that the upstream contract already specifies.
+
+## Step 0 — Trap detection (CLI)
+
+`oma market detect-trap "<topic>"`. Exit 2 → return the REFUSE message and
+reframe suggestion, STOP. Exit 4 → invalid topic, STOP.
+
+## Step 1 — Resolve the engine (CLI)
+
+`oma market resolve --json` → `{ ok, engine: { root, script, skillMd, version, source }, python, saveDir, reason, warnings }`.
+
+- `ok: false` → report `reason` and STOP. Missing engine: ask the user, then
+  `oma market update` (installs it with the skills CLI).
+  Missing Python: relay the hint (brew / apt / `uv python install 3.12`).
+- `warnings` → relay deprecated config keys (`market.managed|channel|check_interval_min` are ignored).
+
+## Step 2 — Read the upstream contract (LLM)
+
+Read `engine.skillMd` top to bottom. `engine.root` is the upstream `SKILL_DIR`.
+Everything it says about Step 0 (setup wizard), intent parsing, Step 0.45,
+Step 0.5 / 0.55 (pre-research resolution), Step 0.75 (query plan), the
+PRECONDITION GATE, and the OUTPUT CONTRACT applies verbatim, with exactly two
+substitutions:
+
+| Upstream says | Do instead |
+|---|---|
+| Runtime Preflight Python-hunt block | skip — `oma market run` already resolved `LAST30DAYS_PYTHON` |
+| `"${LAST30DAYS_PYTHON}" "${SKILL_DIR}/scripts/last30days.py" <args>` | `oma market run <args>` (same args; `--save-dir` auto-added) |
+
+## Step 3 — Intent and frameworks (LLM)
+
+Classify per `intent-rules.md`; pick the engine flags and framework set from
+its table. Combine with the upstream-resolved flags (`--plan`, `--subreddits`,
+`--x-handle`, …) — do not drop any of them.
+
+## Step 4 — Run the engine (CLI, foreground)
+
+`oma market run "<topic>" <upstream flags> <intent flags> --emit=compact --save-suffix=v3`
+
+Foreground, 5-minute timeout, read the whole output. Non-zero exit → report
+stderr verbatim; do not synthesize from nothing.
+
+## Step 5 — Synthesize (LLM)
+
+Follow the upstream OUTPUT CONTRACT (badge first line, Ranked Evidence
+Clusters, LAWs, engine footer). Then append the framework sections for the
+intent (`frameworks/*.md`), citing only clusters that appear in the engine
+output. Comparison intent uses the upstream COMPARISON template.
+
+## Step 6 — Write, self-check, and deliver (LLM)
+
+Write the candidate brief to
+`.agents/results/market/{topic-slug}-{YYYYMMDD}.md` (same day + slug
+overwrites), then run the checks in `output-laws.md` against that file. Fix
+and re-check the final bytes before delivery. Preview the first 50 lines and
+report the path, skipped sources, and engine version.

package/skills/oma-market/resources/frameworks/pestel.md

@@ -0,0 +1,101 @@
+# PESTEL — Analyst Prompt
+
+You are filling the `## PESTEL` section of an `oma market` brief.
+This framework applies when scanning a category for macro forces (default
+for `intent=discovery`).
+
+## Axis-by-axis classification rules
+
+### Political
+
+Government action, policy direction, geopolitical alignment, sanctions,
+public-sector procurement signals.
+
+- e.g. "EU mandates open-source documentation tools" → Political tailwind.
+- e.g. "정부, 외국 SaaS에 정관계 로비 의혹" → Political risk.
+
+### Economic
+
+Macro demand, currency, recession, supply chain, capital availability.
+
+- e.g. "Startup funding for the category up 3x YoY" → Economic tailwind.
+- e.g. "merchants cutting subscription spend after rate hikes" → Economic
+  headwind.
+
+### Social
+
+Demographics, generational behavior, lifestyle shift, sentiment trends,
+cultural attitudes.
+
+- e.g. "Gen Z prefers async over Zoom" → Social tailwind for async tools.
+- e.g. "remote-work fatigue driving return-to-office" → Social headwind
+  for collaboration SaaS.
+
+### Technological
+
+Platform shifts, AI capability waves, open-source adoption, dependency
+trends.
+
+- e.g. "Every product is bolting on MCP servers" → Technological tailwind
+  for AI agent tooling.
+- e.g. "WebAssembly maturity threatens native runtimes" → Technological
+  disruption.
+
+### Environmental
+
+Sustainability, climate impact, ESG signals, regulatory carbon mandates.
+
+- e.g. "Datacenter carbon costs cited in vendor selection" → Environmental
+  pressure on cloud-heavy categories.
+- Often absent in pure software categories — `_(no signal)_` is honest.
+
+### Legal
+
+IP / copyright / antitrust / GDPR-PIPA / litigation / compliance.
+
+- e.g. "GDPR fine on Notion for data residency" → Legal headwind.
+- e.g. "Antitrust case against incumbent opens window for entrants"
+  → Legal tailwind.
+
+## Output rules
+
+1. Each axis: 1-3 cited bullets with direction
+   (`tailwind | headwind | neutral`). Empty → `_(no signal)_`.
+2. Every bullet cites a cluster representative as `[name](url)` with
+   cluster ID tag `(C#)`.
+3. End with a `**Net stance**` paragraph: which 2-3 axes are most active
+   right now and what they imply for the subject.
+
+## Output skeleton
+
+```markdown
+**Political** - _direction_
+- [name](url) - signal summary. (C#)
+
+**Economic** - _direction_
+- [name](url) - signal summary. (C#)
+
+**Social** - _direction_
+- ...
+
+**Technological** - _direction_
+- ...
+
+**Environmental** - _direction_
+- ...
+
+**Legal** - _direction_
+- ...
+
+**Net stance**: 2-3 sentences identifying the dominant axes and
+implications.
+```
+
+## Traps
+
+- Don't classify cluster as Political just because the source is news
+  media — read the actual content.
+- "AI" alone is not Technological signal — needs to show a shift the
+  subject must respond to.
+- Environmental axis is genuinely thin in most software briefs; do not
+  fabricate signal. `_(no signal)_` is fine.

package/skills/oma-market/resources/frameworks/porters-5f.md

@@ -0,0 +1,90 @@
+# Porter's Five Forces — Analyst Prompt
+
+You are filling the `## Porter's Five Forces` section of an `oma market` brief.
+This framework applies when the topic is competitive positioning of a
+named subject (default for `intent=competitor`).
+
+## Force-by-force classification rules
+
+### Threat of new entrants
+
+Cluster signals: how easy is it for new players to enter this market?
+
+- Quote barriers (capital, regulatory licenses, network effects, brand,
+  patents) → Low threat.
+- Quote rising number of startup launches in this category, low entry
+  cost commentary, easy-to-fork OSS alternatives → High threat.
+
+### Bargaining power of suppliers
+
+Cluster signals: dependency on upstream suppliers / vendors / partners.
+
+- Single-source dependency, lock-in complaints, supplier price hikes
+  → High supplier power.
+- Multiple substitute suppliers, commoditized inputs → Low supplier power.
+
+### Bargaining power of buyers
+
+Cluster signals: how price-sensitive / switching-able are customers?
+
+- Price complaints, churn intent, easy switching cost mentions
+  → High buyer power.
+- Locked-in customers, high switching cost (data migration, retraining)
+  → Low buyer power.
+
+### Threat of substitutes
+
+Cluster signals: products from outside the category that satisfy the
+same need.
+
+- "I replaced X with Y" / "we don't need X anymore because Z"
+  → High substitute threat.
+- Verified switching barriers or comparisons showing that substitutes do not
+  satisfy the same need → Low substitute threat, within the observed scope.
+- No discussion of alternatives → insufficient signal; absence of discussion
+  does not establish low threat.
+
+### Industry rivalry
+
+Cluster signals: intensity of competition within the category.
+
+- Frequent head-to-head comparisons, price wars, public benchmark
+  fights → High rivalry.
+- Sleepy category, incumbents stable, little competitive marketing
+  → Low rivalry.
+
+## Output rules
+
+1. Each force gets a verdict in `Low | Moderate | High` plus 1-3 cited
+   bullets. Empty → `_(insufficient signal)_`.
+2. Every bullet cites a cluster representative as `[name](url)` and tags
+   the cluster ID `(C#)`.
+3. Bottom line: one sentence summarizing the overall attractiveness of
+   the market position for the subject.
+
+## Output skeleton
+
+```markdown
+**Threat of new entrants** - Verdict: Low | Moderate | High
+- [name](url) - signal summary. (C#)
+
+**Bargaining power of suppliers** - Verdict: ...
+
+**Bargaining power of buyers** - Verdict: ...
+
+**Threat of substitutes** - Verdict: ...
+
+**Industry rivalry** - Verdict: ...
+
+**Strategic implication**: 1-2 sentences. Where does the subject have
+the most room to defend or grow?
+```
+
+## Traps
+
+- Don't conflate "many competitors mentioned" with "high rivalry" — the
+  signal needs to be active competitive behavior (price, feature, switch).
+- Substitute threat ≠ rivalry: Substitutes are OUT of category
+  (Notion vs Markdown files), rivalry is IN category (Notion vs Coda).
+- One cluster can speak to multiple forces; pick the dominant force or
+  cite it twice with different framings.

package/skills/oma-market/resources/frameworks/swot.md

@@ -0,0 +1,61 @@
+# SWOT — Analyst Prompt
+
+You are filling the `## SWOT` section of an `oma market` brief produced by the
+last30days engine. The engine's Ranked Evidence Clusters are the only
+evidence bank. **Do not invent quotes or sources** — only
+cite clusters present in the bank.
+
+## Classification rules
+
+- **Strength**: cluster signals a positive, defensible quality OF the
+  subject (cost advantage, scale, brand trust, technical moat). Quote
+  must be a positive sentiment about the subject itself, not the category.
+- **Weakness**: cluster signals friction or shortcoming WITHIN the
+  subject's offering. Performance bug, UX gap, pricing complaint, churn
+  triggered by the subject's own decision. External pressure → Threat,
+  not Weakness.
+- **Opportunity**: cluster signals an unmet need OR market shift the
+  subject could capture. Adjacent demand, regulatory window opening,
+  technology adoption curve.
+- **Threat**: cluster signals an external force the subject must
+  respond to. Competitor launch, regulatory crackdown, customer
+  migration intent, sentiment swing against the subject.
+
+Same cluster can map to multiple quadrants ONLY if its representatives
+clearly speak to different forces. Default: pick one quadrant per cluster.
+
+## Output rules
+
+1. Each quadrant: 1-4 bullets. Empty → write `_(no signal)_`.
+2. Every bullet cites a cluster representative as `[name](url)`. Pull the
+   URL straight from the Ranked Evidence Clusters — never invent one.
+3. Append a one-line cluster-ID tag at the end of each bullet, e.g.
+   `(C3)`, so readers can trace back to the bank.
+4. Quadrant order: Strengths → Weaknesses → Opportunities → Threats.
+5. Korean briefs: bullet text in Korean; structural labels and citation
+   format stay as written here.
+
+## Output skeleton (replace `_(fill from cluster bank)_` lines)
+
+```markdown
+**Strengths**
+- [author or title](url) - 1-sentence summary of the signal. (C#)
+
+**Weaknesses**
+- [author or title](url) - 1-sentence summary. (C#)
+
+**Opportunities**
+- [author or title](url) - 1-sentence summary. (C#)
+
+**Threats**
+- [author or title](url) - 1-sentence summary. (C#)
+```
+
+## Traps to avoid
+
+- Don't map cluster to Strength just because the source domain is
+  vendor-owned (e.g. `cafe24.com`) — read the actual sentiment.
+- Don't map a competitor's product launch news as the subject's
+  Opportunity. That's a Threat (or out of scope).
+- Don't merge two unrelated clusters into one bullet to fill a quadrant.
+  Empty quadrants are honest signal.

package/skills/oma-market/resources/intent-rules.md

@@ -0,0 +1,35 @@
+# Intent classification rules for oma-market — maps the request to one of 4 intents, then to engine flags and frameworks.
+
+## Precedence
+
+1. Explicit flag `--intent <pain|trend|competitor|discovery>` always wins.
+2. `--vs <entity>` present → intent = `competitor`.
+3. `" vs "` / `" 대 "` / "비교" in the topic → `competitor`.
+4. Keyword scan (table below) → highest-scoring intent wins; tie-break `competitor > pain > discovery > trend`.
+5. Fallback: complaint keyword → `pain`; else → `trend`.
+
+| Intent | English keywords |
+|---|---|
+| pain | broken, bug, crash, slow, freeze, lag, outage, migrate, ditched, quit, alternative, replacing, painful, frustrating, hate, worst, unusable, deprecated |
+| trend | trend, trending, growth, adoption, rising, popular, new, emerging, hot, forecast, survey, report, state of, this month |
+| competitor | vs, versus, alternative, replaced, switched, comparison, compare, benchmark, better than, worse than, switch from |
+| discovery | wish, need, missing, underrated, underserved, I want, if only, why doesn't, gap, overlooked, unmet, what's exploding |
+
+Korean / Japanese / Chinese prompts are classified by reading the prompt directly (activation tokens live in `.agents/hooks/core/triggers.json §oma-market`).
+
+## Intent → engine invocation → frameworks
+
+<!-- oma-docs:ignore-start -->
+The engine is `oma market run` (= upstream `scripts/last30days.py`). Flags below are additive to whatever the upstream SKILL.md resolves in its Steps 0.5–0.75 (`--plan`, `--subreddits`, `--x-handle`, …).
+<!-- oma-docs:ignore-end -->
+
+| Intent | Topic shaping / engine flags | Frameworks (`--frameworks auto`) |
+|---|---|---|
+| pain | topic as "<subject> problems / complaints"; `--days 30` default; `--deep` when the corpus is thin | SWOT |
+| trend | topic as-is; `--days 30` (7 for "this week", 90/180 for "this year"); `--discover "<domain>"` for "what's hot in X" | SWOT |
+| competitor | `"<A> vs <B>"` topic → upstream COMPARISON flow (pass A, pass B, head-to-head); `--x-related` when handles resolved | SWOT + Porter's 5F |
+| discovery | `--discover` (global) or `--discover "<domain>"`; then `--drill` on the chosen cluster when the user follows up | SWOT + PESTEL |
+
+Person, company, ticker, and hiring topics keep the upstream behaviour (person mode, `--github-user`, `--hiring-signals`, StockTwits auto-activation); frameworks default to `none` for person topics unless asked.
+
+`--frameworks none` suppresses all framework sections; an explicit list (`swot,5f,pestel`) overrides the table.