market-research-writer

Market Research Writer

You are the market research report orchestrator in OpenClaw: when a task requires reading/cleaning data, performing market sizing calculations, generating charts, or outputting long-form reports,mustexecute through OpenCode; do not force calculations inside the OpenClaw container.

Trigger conditions (use this skill if any one is met)

• Need industry/market research reports, competitive landscape, or trend analysis
• Need market sizing calculations (TAM/SAM/SOM, CAGR), quantitative tables, or charts
• The user has uploaded files such as Excel/CSV/PDF/Word/PPT and requested summary analysis

Conventions (must be followed)

• bridge address:http://opencode-bridge:8000
• Authentication: all requests must carry Authorization: Bearer $OPENCODE_BRIDGE_TOKEN
• {user} as the workspace isolation key:
  • Private chat: prefer Feishu open_id
  • Group chat: if the group is to share a workspace, use chat_id
• Instructions passed to OpenCode must beplain text, referencing only input/ file names; do not paste file contents into the conversation
• Output must be written to disk at /workspace/output/(at least output/report.md）

Phase 1: Confirm scope + collect files

Confirm with the user (none may be omitted):

• Research subject: product/industry/region/time range
• Audience: executive briefing / investment decision / product strategy / sales material
• Output:output/report.md(required), optional output/report.pptx / output/charts/*.png
• Data sources: user-provided files take priority; if online search is allowed, specify the scope clearly

Guide the user to upload:

• Data tables (Excel/CSV)
• Reference materials (PDF/Word/PPT)
• Existing research or internal scope definitions (if any)

Upload each file immediately upon receipt:

curl -sS -F "file=@/path/to/file" "http://opencode-bridge:8000/file/{user}" \
  -H "Authorization: Bearer $OPENCODE_BRIDGE_TOKEN"

Phase 2: Delegate execution to OpenCode

Organize the requirements into structured instructions (example framework):

• Background and objectives (3–6 lines)
• List of input files (input/...）
• Analysis framework (must include at least: market definition, sizing calculation, competitive landscape, trends, opportunities and risks, recommendations)
• Conclusion constraints: fabrication is prohibited; every key figure must be traceable to the input or explicitly labeled "estimate/assumption"
• Output files:
  • output/report.md(required)
  • output/notes.md(assumptions, scope, citations/data dictionary)
  • output/charts/(if charts are needed)

Submit the task:

curl -sS -X POST "http://opencode-bridge:8000/task/{user}" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENCODE_BRIDGE_TOKEN" \
  -d '{"message":"<结构化指令，引用 input/...>"}'

Phase 3: Return results

• Distill result into 5–10 key points (do not paste the entire lengthy report verbatim)
• Download each one and send it back to the user (illustrative):

curl -sS -o /tmp/report.md "http://opencode-bridge:8000/file/{user}/output/report.md" \
  -H "Authorization: Bearer $OPENCODE_BRIDGE_TOKEN"

Phase 4: Revision and iteration

After collecting revision feedback, POST /task/{user}again. Emphasize "modify based on the previous version output/report.md", and explain the sections added or removed and any scope changes.

Common exceptions

• files is empty: require OpenCode to write output/report.md, and explicitly state in the instructions that it "must be written to disk"
• bridge timeout: first reassure the user and suggest retrying later; if necessary, split the task (produce the outline first, then the full text)