dse120071750/m8m-harness-builder / examples/text_pipeline
text-pipeline
VerifiedReference M8M 2.0 flowstep_flow_v4 skill with chosen JSON milestone outputs.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 399 B |
| flows/text_pipeline_v1.yaml | config | 2.3 KB |
| planning/flowstep-instruction.md | markdown | 5.1 KB |
| planning/m8m-flowchart.jpg | binary | 146.4 KB |
| planning/m8m-flowchart.md | markdown | 3.4 KB |
| references/ingest.md | markdown | 573 B |
| references/label.md | markdown | 720 B |
| references/segment.md | markdown | 581 B |
| steps/ingest/input.schema.json | config | 398 B |
| steps/ingest/output.schema.json | config | 652 B |
| steps/ingest/tests/test_tool.py | script | 824 B |
| steps/ingest/tool.py | script | 495 B |
| steps/label/draft.schema.json | config | 261 B |
| steps/label/input.schema.json | config | 300 B |
| steps/label/output.schema.json | config | 660 B |
| steps/label/tests/test_tool.py | script | 1.0 KB |
| steps/label/tool.py | script | 902 B |
| steps/segment/input.schema.json | config | 299 B |
| steps/segment/output.schema.json | config | 770 B |
| steps/segment/tests/test_tool.py | script | 769 B |
| steps/segment/tool.py | script | 565 B |
planning/m8m-flowchart.md
View on GitHub# M8M flowchart: text_pipeline_v1
One chart. Milestone to milestone. Each node declares named output ports.
FlowSteps refine candidates; admission always runs and an authored judge is optional. Without chosen-output.json the node is BLOCKED.
Run state is authoritative. Optional cross-run cache only supplies a candidate for the current judge.
Execution context is isolated: a fresh run does not inherit orchestration chat or prior-row memory.
FlowSteps inside a node are a guide (one preferred tool each), not a compulsory path.
cycle = wrap over a frozen ledger (pass preserves, fail purges). judge = optional semantic retry until PASS.
The JPEG is the audit copy: portable, human-labeled, native to review.
It is rewritten on generate and on every step edit.
- flow_id: `text_pipeline_v1`
- source: `edit`
- updated_at: 2026-09-20T14:04:28Z
## Chart
Portable JPEG for audit. Humanizer names each milestone and FlowStep.
Regenerated on generate and on every step edit (`write` / `mark`).

```text
flowchart TD
request([request])
ingest["ingest<br/>out:result"]
segment["segment<br/>out:result"]
label["label<br/>intel:completion<br/>out:result"]
request --> ingest
ingest --> segment
segment --> label
```
## Toolbox plan
Tools on each proposed milestone. **Existing toolbox** = already in
`<repo>/flowsteps/tools/` or an M8M seed. **Promote from a skill script** =
skill-private Python becomes that tool. **Generate new** = builder should
develop this tool; a stub is a successful sketch.
| Milestone | Intelligence | Existing toolbox | Promote from a skill script | Generate new |
| --- | --- | --- | --- | --- |
| `ingest` | `none` | — | — | — |
| `segment` | `none` | — | — | — |
| `label` | `completion` (semantic class is not derivable from punctuation alone) | — | — | — |
## Nodes
| Milestone | What it means | Success | Declared output ports | Status | Intelligence | Tools | Control | Cache |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| `ingest` | Ingest | Normalized text is available as the chosen JSON result. | `result` (json, one, required) | `DONE` | `none` | none | linear | off |
| `segment` | Segment | The sentence array is available as the chosen JSON result. | `result` (json, one, required) | `DONE` | `none` | none | linear | off |
| `label` | Label | The first sentence has one accepted label. | `result` (json, one, required) | `DONE` | `completion` | none | linear | off |
## FlowSteps (guide)
Sequence inside each milestone. Prefer the named tool. Optional.
Start from the complete milestone master prompt, including during recovery.
FlowStep notes are optional. Admission is compulsory; semantic judgment is optional.
| Milestone | # | FlowStep | What it means | Preferred tool | Milestone master prompt |
| --- | ---: | --- | --- | --- | --- |
| (none) | | | | | |
## Cycle
None. No cycle wrap. Freeze a ledger first, then wrap milestones.
## Optional semantic judge
None. These milestones commit after structural admission with zero judge calls.
## Branch (after the milestone)
None. No branch after a milestone.
Proceed only after structural admission and, when declared, semantic judge `PASS`; then commit current named outputs as `chosen-output.json`.
Branch is after that commit. Candidate handlers never write control; the runtime-owned control call writes `branch`.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |