[SUPERSEDED][AP0][MIGRATION] Bundle AgentProlog + harness in prolog-rlm and salvage prior Harness work

lost-rob0t commented

2026-08-25 02:13:05 +00:00

(Migrated from github.com)

Current architectural decision — 2026-08-27

This issue is the canonical migration track after the user correction in this thread.

lost-rob0t/prolog-rlm is the source repository for both the reusable runtime and the bundled reference AgentProlog/harness application. We are not maintaining a separate downstream AgentProlog product or a separate DeepSeek-Harness product track.

PR #132 is the current implementation base. The older PR #125 / official-DSH integration is a salvage source, not a branch or product architecture to revive.

Target repository boundary

lost-rob0t/prolog-rlm
├── prolog/                 reusable SWI-Prolog RLM runtime/library
├── agentProlog/            bundled application/config composition
├── harness/                bundled reference frontend/harnesses
├── tools/ / packs/         separately loadable concrete tool packages
└── flake.nix               library + application + harness outputs

Dependency direction remains strict:

harness -> agentProlog application facade -> public prolog-rlm APIs

Reusable prolog/ code must not import frontend/product code. Shipping everything from one repository does not mean collapsing the library/application boundary.

Superseded architecture

The following are no longer the target:

standalone lost-rob0t/agentProlog as the product owner;
prolog-rlm acting only as a downstream-consumed flake runtime;
#184 as an optional independent official DeepSeek Harness / AgentFactory product integration;
reviving PR #125 wholesale;
a DSH submodule/source checkout as a required runtime/build dependency;
a separate deepseek-harness.json settings authority;
a second provider, history, tool, authority, compaction, planning, recursion, or verification runtime.

Salvage from PR #125 / prior Harness work

Migrate useful semantics into the in-repo application/harness only when they still fit the canonical runtime boundary.

Prior work	Action
persistent bridge/process lifecycle ideas	SALVAGE where needed
explicit handshake/versioning	SALVAGE
cancellation reaching canonical Prolog execution	SALVAGE
bridge crash -> terminal structured failure	SALVAGE
request/session correlation + duplicate suppression tests	SALVAGE
focused harness CI / integration fixtures	SALVAGE
upstream DSH submodule and full monorepo build	DROP
custom DSH headless runner	DROP
broad Cordis/profile disable patches	DROP
private DeepSeek settings store	DROP
Harness-owned provider/tool/history semantics	DROP

Do not mechanically copy old files. Re-implement the useful contracts against current main and PR #132 APIs with TDD.

Canonical ownership

Prolog-RLM remains authoritative for:

providers/model calls and usage;
prompt/context compilation and permanent skills;
tools, capabilities, authority and durable effects;
conversations/history/memory;
planning, recursion/subagents and graph execution;
source/project KB, retrieval and freshness;
Spec/Plan/Verify/repair;
tracing and canonical status.

The bundled application/harness owns presentation, operator configuration, process adaptation and ergonomics only.

Current implementation base

PR #132 already moves the product/application back into this repository:

harness/deepseek_tui
        ↓
agentProlog/
        ↓
prolog/

Continue from that current-main recovery branch rather than PR #125.

Next implementation slice — TDD first

Add a renderer-independent in-repo harness transport contract around the canonical AgentProlog/RLM execution path instead of treating one-shot CLI stdout as the final UI protocol.
Prove request identity/correlation and exactly-once projection for a user turn/result.
Add real cancellation propagation from harness -> canonical Prolog runtime.
Add explicit bridge/process crash and protocol-version failure behavior.
Preserve canonical runtime status/usage projection, including model, token input/output, and context occupancy/unknown.
Keep the current simple TUI as a consumer of that contract; do not move semantics into Go/frontend code.
Port only focused regression tests/CI concepts from #125 that prove these contracts.

Acceptance gate

one repository contains the reusable library, AgentProlog application and bundled harness;
clean prolog-rlm library/flake/SWI-pack consumption still works independently;
no separate downstream AgentProlog product is required;
no official DSH repo/submodule/build is required for the bundled harness;
one harness turn maps to one canonical Prolog-RLM execution trajectory;
cancellation reaches canonical runtime execution;
bridge/process failure is explicit and terminal, never fake success;
request/result correlation is deterministic and duplicate-safe;
config remains one canonical AgentProlog/Prolog configuration authority;
frontend code does not own providers/tools/history/planning/authority/verification;
useful PR #125 regression coverage is migrated without reviving its architecture;
focused + aggregate deterministic tests and Nix checks remain green;
live-provider lanes remain separately gated and do not weaken deterministic CI.

#141 is superseded because it encodes standalone downstream AgentProlog ownership.
#231's previous closing rationale is obsolete; useful generic runtime-status work remains valid, while this issue now owns the bundled-product/harness migration.
PR #132 is the active implementation base.

Do not reintroduce the superseded split in later issue/PR comments.

## Current architectural decision — 2026-08-27 This issue is the canonical migration track after the user correction in this thread. **`lost-rob0t/prolog-rlm` is the source repository for both the reusable runtime and the bundled reference AgentProlog/harness application. We are not maintaining a separate downstream AgentProlog product or a separate DeepSeek-Harness product track.** PR #132 is the current implementation base. The older PR #125 / official-DSH integration is a **salvage source**, not a branch or product architecture to revive. ## Target repository boundary ```text lost-rob0t/prolog-rlm ├── prolog/ reusable SWI-Prolog RLM runtime/library ├── agentProlog/ bundled application/config composition ├── harness/ bundled reference frontend/harnesses ├── tools/ / packs/ separately loadable concrete tool packages └── flake.nix library + application + harness outputs ``` Dependency direction remains strict: ```text harness -> agentProlog application facade -> public prolog-rlm APIs ``` Reusable `prolog/` code must not import frontend/product code. Shipping everything from one repository does **not** mean collapsing the library/application boundary. ## Superseded architecture The following are no longer the target: - standalone `lost-rob0t/agentProlog` as the product owner; - `prolog-rlm` acting only as a downstream-consumed flake runtime; - #184 as an optional independent official DeepSeek Harness / AgentFactory product integration; - reviving PR #125 wholesale; - a DSH submodule/source checkout as a required runtime/build dependency; - a separate `deepseek-harness.json` settings authority; - a second provider, history, tool, authority, compaction, planning, recursion, or verification runtime. ## Salvage from PR #125 / prior Harness work Migrate useful semantics into the in-repo application/harness only when they still fit the canonical runtime boundary. | Prior work | Action | |---|---| | persistent bridge/process lifecycle ideas | SALVAGE where needed | | explicit handshake/versioning | SALVAGE | | cancellation reaching canonical Prolog execution | SALVAGE | | bridge crash -> terminal structured failure | SALVAGE | | request/session correlation + duplicate suppression tests | SALVAGE | | focused harness CI / integration fixtures | SALVAGE | | upstream DSH submodule and full monorepo build | DROP | | custom DSH headless runner | DROP | | broad Cordis/profile disable patches | DROP | | private DeepSeek settings store | DROP | | Harness-owned provider/tool/history semantics | DROP | Do not mechanically copy old files. Re-implement the useful contracts against current `main` and PR #132 APIs with TDD. ## Canonical ownership Prolog-RLM remains authoritative for: - providers/model calls and usage; - prompt/context compilation and permanent skills; - tools, capabilities, authority and durable effects; - conversations/history/memory; - planning, recursion/subagents and graph execution; - source/project KB, retrieval and freshness; - Spec/Plan/Verify/repair; - tracing and canonical status. The bundled application/harness owns presentation, operator configuration, process adaptation and ergonomics only. ## Current implementation base PR #132 already moves the product/application back into this repository: ```text harness/deepseek_tui ↓ agentProlog/ ↓ prolog/ ``` Continue from that current-main recovery branch rather than PR #125. ## Next implementation slice — TDD first 1. Add a renderer-independent in-repo harness transport contract around the canonical AgentProlog/RLM execution path instead of treating one-shot CLI stdout as the final UI protocol. 2. Prove request identity/correlation and exactly-once projection for a user turn/result. 3. Add real cancellation propagation from harness -> canonical Prolog runtime. 4. Add explicit bridge/process crash and protocol-version failure behavior. 5. Preserve canonical runtime status/usage projection, including model, token input/output, and context occupancy/unknown. 6. Keep the current simple TUI as a consumer of that contract; do not move semantics into Go/frontend code. 7. Port only focused regression tests/CI concepts from #125 that prove these contracts. ## Acceptance gate - [ ] one repository contains the reusable library, AgentProlog application and bundled harness; - [ ] clean `prolog-rlm` library/flake/SWI-pack consumption still works independently; - [ ] no separate downstream AgentProlog product is required; - [ ] no official DSH repo/submodule/build is required for the bundled harness; - [ ] one harness turn maps to one canonical Prolog-RLM execution trajectory; - [ ] cancellation reaches canonical runtime execution; - [ ] bridge/process failure is explicit and terminal, never fake success; - [ ] request/result correlation is deterministic and duplicate-safe; - [ ] config remains one canonical AgentProlog/Prolog configuration authority; - [ ] frontend code does not own providers/tools/history/planning/authority/verification; - [ ] useful PR #125 regression coverage is migrated without reviving its architecture; - [ ] focused + aggregate deterministic tests and Nix checks remain green; - [ ] live-provider lanes remain separately gated and do not weaken deterministic CI. ## Related state cleanup - #141 is superseded because it encodes standalone downstream AgentProlog ownership. - #231's previous closing rationale is obsolete; useful generic runtime-status work remains valid, while this issue now owns the bundled-product/harness migration. - PR #132 is the active implementation base. Do not reintroduce the superseded split in later issue/PR comments.

lost-rob0t commented

2026-08-25 02:42:23 +00:00

(Migrated from github.com)

approved for design

lost-rob0t commented

2026-08-25 03:10:33 +00:00

(Migrated from github.com)

Approval reconciliation

The design was explicitly approved by the user in this issue thread (approved for design). The title has been updated to [ADARD][APPROVED].

Implementation ordering is now:

complete #186 repository cleanup and removal of legacy nested/product harnesses;
then implement this approved plugin-only Harness architecture from the cleaned runtime boundary.

The old PR #125/fork-style Harness path is not the implementation base for this ADARD.

## Approval reconciliation The design was explicitly approved by the user in this issue thread (`approved for design`). The title has been updated to `[ADARD][APPROVED]`. Implementation ordering is now: 1. complete #186 repository cleanup and removal of legacy nested/product harnesses; 2. then implement this approved plugin-only Harness architecture from the cleaned runtime boundary. The old PR #125/fork-style Harness path is not the implementation base for this ADARD.

lost-rob0t commented

2026-08-25 04:50:58 +00:00

(Migrated from github.com)

Current workflow status

The ADARD design is approved by the user (see the existing approved for design comment). The stale pre-approval sentence in the original issue body is superseded by that approval and the current issue title.

Implementation ordering is now:

land #186 / PR #187 and recheck the runtime-only main boundary;
begin the approved #184 plugin-only Harness core from that clean base;
return to a design gate only if implementation evidence requires changing an approved architectural boundary.

Do not revive #124 / PR #125 or a nested agentProlog/ harness as the implementation base.

## Current workflow status The ADARD design is **approved** by the user (see the existing `approved for design` comment). The stale pre-approval sentence in the original issue body is superseded by that approval and the current issue title. Implementation ordering is now: 1. land #186 / PR #187 and recheck the runtime-only `main` boundary; 2. begin the approved #184 plugin-only Harness core from that clean base; 3. return to a design gate only if implementation evidence requires changing an approved architectural boundary. Do not revive #124 / PR #125 or a nested `agentProlog/` harness as the implementation base.

lost-rob0t commented

2026-08-26 09:01:48 +00:00

(Migrated from github.com)

Architecture clarification approved in #231: DeepSeek Harness is the reference IDE/workspace renderer for the bundled prolog-rlm CLI, not the semantic agent runtime and not a separate AgentProlog product owner. Prolog-RLM keeps canonical ownership of context compilation/projection, provider/model calls, tools/capabilities/authority/effects, history/memory, recursion/subagents, project/source KB, retrieval, verification/repair, and usage/status. The DSH adapter should consume prolog_agent_ui_v1; if that protocol lacks required semantics, extend/version the protocol rather than importing runtime internals into DSH. Keep a one-shot prolog-rlm run "task" path alongside prolog-rlm ide .. Canonical status includes model + token input/output + context occupancy percent when known (ctx ? when capacity is unknown). #233 is the first TDD implementation slice.

Architecture clarification approved in #231: DeepSeek Harness is the **reference IDE/workspace renderer for the bundled `prolog-rlm` CLI**, not the semantic agent runtime and not a separate AgentProlog product owner. Prolog-RLM keeps canonical ownership of context compilation/projection, provider/model calls, tools/capabilities/authority/effects, history/memory, recursion/subagents, project/source KB, retrieval, verification/repair, and usage/status. The DSH adapter should consume `prolog_agent_ui_v1`; if that protocol lacks required semantics, extend/version the protocol rather than importing runtime internals into DSH. Keep a one-shot `prolog-rlm run "task"` path alongside `prolog-rlm ide .`. Canonical status includes model + token input/output + context occupancy percent when known (`ctx ?` when capacity is unknown). #233 is the first TDD implementation slice.

lost-rob0t commented

2026-08-27 13:55:07 +00:00

(Migrated from github.com)

Scope clarification — 2026-08-27

PR #132 now adds a small in-repo DeepSeek API harness with a Bubble Tea v2 TUI over the AgentProlog CLI. That does not silently implement or replace this issue's separate official DeepSeek Harness / AgentFactory-plugin integration design.

Current split:

#132: AgentProlog monorepo application layer + direct DeepSeek provider profile + Bubble Tea/Bubbles terminal frontend;
#184: optional future integration with the official deepseek-harness product/plugin architecture, if still wanted.

Both must route semantic execution through Prolog-RLM and neither may create a second settings/provider/authority runtime. Do not revive the old #125 nested Harness implementation as part of #132.

## Scope clarification — 2026-08-27 PR #132 now adds a **small in-repo DeepSeek API harness with a Bubble Tea v2 TUI** over the AgentProlog CLI. That does **not** silently implement or replace this issue's separate official DeepSeek Harness / AgentFactory-plugin integration design. Current split: - #132: AgentProlog monorepo application layer + direct DeepSeek provider profile + Bubble Tea/Bubbles terminal frontend; - #184: optional future integration with the official `deepseek-harness` product/plugin architecture, if still wanted. Both must route semantic execution through Prolog-RLM and neither may create a second settings/provider/authority runtime. Do not revive the old #125 nested Harness implementation as part of #132.

lost-rob0t commented

2026-08-27 13:57:20 +00:00

(Migrated from github.com)

No GPT, it does replace it because we're not gonna use that repo. Everything should be provided from here. It's so simple. I can't believe you can't see it. Of course. To harness work will need to be migrated back to it.

nsaspy referenced this issue

2026-08-27 20:02:08 +00:00

Bundle AgentProlog + DeepSeek harness in prolog-rlm #132

nsaspy referenced this issue

2026-08-27 20:02:14 +00:00

Cleanup legacy nested harnesses before #184 #187

nsaspy referenced this issue

2026-08-27 20:02:17 +00:00

feat: make prolog-rlm the reference harness with DSH IDE boundary #233

nsaspy referenced this issue

2026-08-27 20:02:19 +00:00

feat: salvage renderer-neutral runtime status projection #263

nsaspy referenced this issue

2026-09-02 17:26:02 +00:00

[CANONICAL] Standalone AgentProlog downstream product boundary #333

nsaspy referenced this issue

2026-09-02 17:26:07 +00:00

[MOVED DOWNSTREAM] DeepSeek Harness integration belongs in lost-rob0t/agentProlog #332

nsaspy referenced this issue

2026-09-02 17:26:08 +00:00

[MOVED DOWNSTREAM] DeepSeek Harness integration belongs in lost-rob0t/agentProlog #332

nsaspy referenced this issue

2026-09-02 17:27:18 +00:00

Bundle AgentProlog + DeepSeek harness in prolog-rlm #132

nsaspy referenced this issue

2026-09-08 01:29:00 +00:00

[SUPERSEDED][AP0] Make prolog-rlm the reference harness with DeepSeek Harness IDE #231

nsaspy changed title from ~~[AP0][MIGRATION] Bundle AgentProlog + harness in prolog-rlm and salvage prior Harness work~~ to [SUPERSEDED][AP0][MIGRATION] Bundle AgentProlog + harness in prolog-rlm and salvage prior Harness work

2026-09-08 01:29:03 +00:00

nsaspy closed this issue

2026-09-08 01:29:03 +00:00

nsaspy commented

2026-09-08 01:44:51 +00:00

Owner

Closing as outdated/superseded architecture. The canonical repository boundary is now #141: lost-rob0t/agentProlog owns the standalone product and prolog-rlm remains the reusable domain-neutral runtime. The bundled AgentProlog/harness ownership described here must not be reintroduced.

Closing as **outdated/superseded architecture**. The canonical repository boundary is now #141: `lost-rob0t/agentProlog` owns the standalone product and `prolog-rlm` remains the reusable domain-neutral runtime. The bundled AgentProlog/harness ownership described here must not be reintroduced.

Rows
Columns

[SUPERSEDED][AP0][MIGRATION] Bundle AgentProlog + harness in prolog-rlm and salvage prior Harness work #184

Current architectural decision — 2026-08-27

Target repository boundary

Superseded architecture

Salvage from PR #125 / prior Harness work

Canonical ownership

Current implementation base

Next implementation slice — TDD first

Acceptance gate

Related state cleanup

Approval reconciliation

Current workflow status

Scope clarification — 2026-08-27