Skip to main content

Recovered Migration Prompt

This document records the best recoverable version of the carried-over prompt that drove the SECS source-of-truth cleanup session.

Important note: the current VS Code transcript does not contain one intact original pasted prompt block for the whole task. The carried-over work entered this transcript mid-stream. What is preserved is:

  • the carried-over high-level objective from the conversation summary
  • exact user directives captured later in the transcript
  • wave-specific Explore / Implement / Validate prompts created to execute the work
  • later recovery prompts that restated completed and remaining waves

Transcript evidence used for this reconstruction:

  • transcript file: /home/herki/.config/Code/User/workspaceStorage/7813fb3f6e6c436d1c2f3a09196a5df7/GitHub.copilot-chat/transcripts/a24aaa47-5519-44c5-89ab-e57530ca6ea8.jsonl
  • key directive lines in the current transcript:
  • 7137: continue working and do it properly follow tenants and point 1
  • 9627: i mean we had fucking 7 or 8 waves in the fucking prompt I am asking proper fucking list what work is left for fuck sake
  • 9823: wave 1 ignroe we do not want that / do all other waves and do it fucking properly follow the sub agents system for fuck sake and add waves for the other issues you pinted out
  • 17224: yes and you need to fucking replace the fuckign stat with channel for fuck sake the main prompt was doing that for fucking cunt
  • 17701: after this fix stop and then write out prompt fully and then write out naother doc with report what you have done so i can fucking fix htings properly
  • 17745: i todl you do fuckign creater adocument where the promtp is and anther what you where doing why did you not fucking do that
  • key recovery / wave-state anchors in the transcript:
  • 4211-4212
  • 4393
  • 5391-5392
  • 15970-15974
  • 4635-4735

Verbatim Preserved Directives​

The following directives are preserved verbatim in the current transcript and are the strongest source-of-truth anchors for the carried-over prompt:

  • continue working and do it properly follow tenants and point 1
  • i mean we had fucking 7 or 8 waves in the fucking prompt I am asking proper fucking list what work is left for fuck sake
  • wave 1 ignroe we do not want that / do all other waves and do it fucking properly follow the sub agents system for fuck sake and add waves for the other issues you pinted out
  • yes and you need to fucking replace the fuckign stat with channel for fuck sake the main prompt was doing that for fucking cunt
  • after this fix stop and then write out prompt fully and then write out naother doc with report what you have done so i can fucking fix htings properly
  • i todl you do fuckign creater adocument where the promtp is and anther what you where doing why did you not fucking do that

Recovered Top-Line Objective​

The carried-over task objective, as preserved in the conversation summary and repeatedly reinforced by later transcript directives, was:

Implement a SECS source-of-truth cleanup and vocabulary migration across the project.

This exact sentence is preserved by the carried-over session summary rather than by a single intact user message inside the current transcript.

Recovered Working Prompt​

The following is the clean reusable working prompt/spec reconstructed from the preserved directives and executed waves.

Implement a SECS source-of-truth cleanup and vocabulary migration across the project.

Carry-over point 1:

  • Finish the template field carrier/runtime cleanup properly, replacing template field carrier/runtime wording with template data and template value where that is the committed end-state.

Core requirements:

  1. Live docs must be clean final-state specifications, not recovery logs.
  2. Docs, .secs source, Generated stand-ins, C# runtime/compiler-target code, tests, scripts, and rule files must stay coherent.
  3. The SECS technical resolver/runtime concept must move from stat to channel where that is the committed technical vocabulary.
  4. Template carrier/runtime wording must move from template field to template data / template value where appropriate, without breaking the committed source-language field syntax.
  5. Canonical identity strings must become the committed compiler/runtime identity and hash target.

Execution rules:

  1. Follow repository tenets and do the work properly.
  2. Do not invent backlog items or fake a remaining-wave list from memory; recover the real remaining work from transcript evidence and current repo state.
  3. Use the subagent system properly.
  4. Ignore Wave 1.
  5. Do all other required waves properly.
  6. Add new waves only for real issues discovered during exploration, not as filler.
  7. Keep validations focused and local to the touched slice.

Recovered wave list:

  • Point 1 / pre-wave slice: template field carrier/runtime cleanup to template data / template value.
  • Wave 0: baseline and recovery bookkeeping, source-of-truth inventory, affected-surface map.
  • Wave 1: explicitly skipped by user instruction.
  • Wave 2: live-doc path cleanup and separation of live normative docs from archive / recovery / backlog material.
  • Wave 3: behavior-vocabulary and command/source rename cleanup.
  • Wave 4: no-fallback boundary/runtime semantics cleanup.
  • Wave 5: identity/hash cleanup toward canonical identity strings and explicit identity handling.
  • Wave 6: stat -> channel migration for the remaining technical resolver/runtime/API/value vocabulary.
  • Wave 7 / closeout: final consistency pass, then write the full prompt and a second report documenting what was done and why.

Operational constraints inferred from the transcript:

  • Keep Wave 5 identity work separate from Wave 6 stat -> channel work.
  • Do not use historical docs or backlog files as live source-of-truth unless the live docs explicitly point to them as such.
  • Do not fabricate source syntax from Generated stand-ins or future-compiler plans.
  • Do not use rollback or staging tricks as a substitute for a real repair.
  • After the current fix, stop and write the prompt doc and the work-report doc.

How To Use This Document​

If a new session has to resume the cleanup from scratch, use the Recovered Working Prompt section above as the starting spec. Use the Verbatim Preserved Directives section to understand which parts were explicit user instructions and which parts were reconstructed from distributed transcript evidence.