← Corpus / dididecks-ai / agent-skill

dididecks-ai/agent-skills/pseudomonorepos/references/lifecycle-workflow

Path
agent-skills/pseudomonorepos/references/lifecycle-workflow.md

The 5-Phase Lifecycle Workflow

The canonical loop for taking a unit of work from idea → ship → reflection → public artifact in the Lossless ecosystem. A diagram and rationale for what will become its own skill.

Status: Working sketch. The full skill (working title lossless-loop) will encode automation, templates, and verification. For now: this doc is the spec.

Universal directories

Every repo level — pseudomonorepo, true monorepo, project repo — should have both:

<any-repo>/
├── context-v/        # living documentation
└── changelog/        # ship log (dated records of change)

These are siblings at the repo root. Some legacy projects nest changelog inside context-v (context-v/changelog/); aspiration is parallel.

The loop

┌──────────┐    ┌──────────┐    ┌──────────┐    ┌──────────┐    ┌──────────┐
│  START   │ →  │ PROGRESS │ →  │ REFLECT  │ →  │ PUBLISH  │ →  │  MARKET  │
│          │    │          │    │          │    │          │    │          │
│ project/ │    │ project/ │    │ parent/  │    │ parent/  │    │ astro-   │
│ context- │    │ change-  │    │ context- │    │ change-  │    │ knots    │
│ v/       │    │ log/     │    │ v/       │    │ log/     │    │ site     │
└──────────┘    └──────────┘    └──────────┘    └──────────┘    └──────────┘
     ↑                                                                  │
     └──────────────── feedback / next iteration ──────────────────────┘
flowchart LR
    A["<b>START</b><br/>project/context-v/<br/><i>spec, prompt, blueprint</i>"]
    B["<b>PROGRESS</b><br/>project/changelog/<br/><i>what shipped, when</i>"]
    C["<b>REFLECT</b><br/>parent/context-v/<br/><i>cross-cutting blueprint, lessons</i>"]
    D["<b>PUBLISH</b><br/>parent/changelog/<br/><i>announce across the family</i>"]
    E["<b>MARKET</b><br/>astro-knots site<br/><i>essay, gallery, walkthrough</i>"]
    A --> B --> C --> D --> E
    E -.->|feedback| A

Phase 1 — Start

Where: project/context-v/ (specs, prompts, blueprints)

What: Frame the work before doing it.

  • Spec: what is being built and why
  • Prompt: step-by-step plan referencing the spec
  • Blueprint: pattern this work will follow (if known)

Output: at least one document the agent can load as context for the build.

Phase 2 — Progress

Where: project/changelog/

What: Log what shipped and when.

  • Dated entries (e.g., 2026-05-03_01.md)
  • Each entry links back to the spec/prompt it implemented
  • Brief, factual: what changed, why, what’s next

Output: a trail anyone (human or AI) can read to understand the project’s evolution.

Phase 3 — Reflect

Where: parent-pseudomonorepo/context-v/ (often blueprints, sometimes specs or explorations)

What: Lift learnings to the cross-cutting level.

  • “We solved X this way in project-a — should it be a blueprint for project-b and project-c?”
  • “We hit Y in three projects — time for a reminder”
  • “Our approach to Z is starting to consolidate — write it up as a blueprint”

Output: parent-level documents that aggregate insight from individual projects.

Phase 4 — Publish

Where: parent-pseudomonorepo/changelog/

What: Announce the change at the family level.

  • “New blueprint added: X”
  • “All Astro Knots sites updated to pattern Y”
  • “Pseudomonorepo-wide convention: Z”

Output: family-level visibility of a project-level change.

Phase 5 — Market

Where: wherever it fits on an Astro Knots site

What: Public-facing artifact.

  • An essay on lossless.group
  • A gallery entry under /projects/gallery/
  • A walkthrough or “Lost in Public” entry
  • A talk, a demo, a Loom — whatever the work calls for

Output: something an outsider can encounter, learn from, and link to.

When to skip phases

Not every unit of work runs all five. Heuristics:

Type of workRealistic phases
Typo / one-line fixJust Progress, maybe
Bug fixStart (issue) → Progress
New feature in one projectStart → Progress
Pattern that affects 2+ projectsStart → Progress → Reflect
Significant new conventionStart → Progress → Reflect → Publish
Any of the above worth public discussionAll five

Skipping is allowed; forgetting is not. When you skip, log the skip as refactor debt — same pattern as the ship-fast escape hatch in search-first.md.

Composition with other skills

This phase…Activates these skills
Startcontext-vigilance (writing in context-v/), pseudomonorepos (search-first)
Progresscontext-vigilance (changelog conventions, TBD)
Reflectcontext-vigilance (writing blueprints), pseudomonorepos (parent-level work)
Publishcontext-vigilance (changelog)
Marketastro-knots (publishing), lfm forthcoming (content authoring)

The loop is the spine; the other skills are the muscles.

TBDs for the forthcoming lossless-loop skill

When this becomes its own skill, it should encode:

  • Changelog entry frontmatter and naming convention (currently looks like YYYY-MM-DD_NN.md)
  • Templates for each phase
  • Automation: a /loop start command? A /loop progress to log a shipment?
  • Verification: “have you reflected this work upward?”
  • Cross-references: how Marketing-phase artifacts link back to Reflection-phase blueprints
  • Where Marketing lives by site (gallery vs. essay vs. lost-in-public)