Independent use will determine whether BDFL’s versioned planning, human approval, isolated worker execution, and recovery workflow reliably coordinates Codex and Claude Code on real software projects.

state: expiredheat: lowuncertainty: highknownscott: mediumagent-harnesses multi-agent-orchestration coding-agentsBDFL

What is this?

BDFL is presented in the supplied evidence titles as an open-source supervisor for coordinating Codex and Claude Code, with an initial implementation committed to a plugin repository. The hypothesis attributes versioned planning, human approval, isolated worker execution, and recovery workflows to it, but the web snippets do not independently document those features, identify who BDFL is, or report real-world adoption. The snippets only establish relevant context: Codex and Claude Code have different execution environments, while practitioners emphasize structured planning, project instructions, and tool-independent workflows.

Why it matters to Scott

The architecture is already held in Scott’s Markdown OS, Long-Running Agents, and Deterministic Agent Control Plane pages, and is implemented in Proposal Compiler using fresh Claude/Codex workers, durable state, recovery, and explicit workflow control. BDFL could provide a directly comparable open-source implementation and eventual independent validation, but the supplied evidence does not yet establish its claimed mechanics, adoption, or reliability.

ip:framework.markdown-osip:framework.long-running-agentsip:source.designing-loops-not-promptsdev:concept.deterministic-agent-control-planedev:concept.resumable-agent-job-control-planedev:project.proposalradar:concept.agent-harnessesradar:concept.coding-agentsradar:concept.multi-agent-coordinationradar:concept.agent-sandboxing

queries asked of Scott's wikis

versioned plans as agent coordination state
human approval gates for coding agents
isolated worker execution and sandboxing
recovery and resumability in agent harnesses
model-agnostic coding-agent orchestration
supervisor-worker patterns for software agents

Measured heat

no measured readings yet — the hourly heat pass fills this in

How the heat travelled

no chain yet — the hourly chain pass fills this in

Evidence (12) — ⭐ canonical anchor

source	object	author	score	comments
🟧 hn	Show HN: BDFL – Open-source supervisor for Codex and Claude Code	thisisnsh	1	3
🟧 echo.github ⭐	The repository’s first commit, “Initialize the BDFL plugin repository,” contains the original BDFL implementation. Its package description c	Nishant Hada	—	—
🟠 reddit	Moving Execution Authority Out of the Model LocalLLaMA	Jay299792458	0	5
🟠 reddit	Autonomous agents are the easy part. The hard part was the queue that runs them. ClaudeAI	ma8nk	1	1
🟠 reddit	The one habit that made Claude Code actually trustworthy on big multi-file changes ClaudeAI	Colin_Pepin	0	10
🟠 reddit	I let agents write every line of my product. Two weeks of building gates came first. ClaudeAI	thebvg	1	1
🟠 reddit	I let Claude build itself (387 PRs later): a crash-only harness where you swap the worker model and the orchestrator model independently ClaudeAI	rehtorical	0	11
🟧 hn	Show HN: Dn – plan collaboratively, let agents execute	moochmooch	2	0
🟠 reddit	Claude Code Has Subagents. Should Anthropic Add a Native Dependency-Aware WBS? ClaudeAI	RFOK	1	2
🟧 hn	A Hermes Agent Skill Looping Between Codex and Claude Code	yruzin	1	0
🟧 hn	Show HN: Handoff is await human() for AI agents	LivingGlitcher	1	1
🟠 reddit	I've been running Claude Code as the orchestrator for a fleet of other agents for about five months. Open sourced the whole thing today (MIT) ClaudeAI	AccomplishedLab3697	0	35

Interpretation history

2026-08-02T22:22:20Z
Repeated parallel implementations now establish the broader supervisor-workflow pattern, but sustained monitoring has produced no independent BDFL adopter, feature verification, or real-project outcome. Retire this product-specific episode unless direct BDFL usage evidence emerges.
2026-08-02T21:21:45Z
The five-month o8 deployment is the strongest practical validation yet of the broader supervisor pattern, including isolated worktrees, shared coordination, and human-approved integration. It still does not establish independent BDFL use or test BDFL’s recovery and cross-model reliability, so the product-specific hypothesis remains open.
2026-08-02T21:21:09Z
evidence attached: reddit.post.1vdtnhv — Independent open-source use of an orchestrator, isolated worktrees, shared coordination, and human-approved merges directly tests the supervisor workflow hypothesis.
2026-08-02T13:22:37Z
The await-human primitive adds another independent implementation of the broader approval-gate pattern, but provides no evidence of BDFL adoption, feature reliability, or real-project outcomes. Product-level validation remains absent and further generic pattern examples should not raise attention.
2026-08-02T13:21:15Z
evidence attached: hn.story.49143842 — The await-human primitive materially contextualizes the open question of whether explicit human approval and recovery workflows make multi-agent coding harnesses reliable.
2026-08-01T17:25:47Z
The Codex–Claude loop is another weak parallel implementation of cross-agent orchestration, but it supplies no independent BDFL adoption, feature verification, or measured reliability result. Pattern corroboration is now repetitive; the product-specific case should remain cold until a substantive user report appears.
2026-08-01T17:21:38Z
evidence attached: hn.story.49136162 — Independent implementation of a Codex-and-Claude-Code autonomous loop materially bears on whether cross-agent coding orchestration works in practice.
2026-07-29T19:25:28Z
The dependency-aware WBS report adds a small practical example of scoped, checked multi-agent coordination, but still does not demonstrate independent BDFL use or validate its integrated recovery workflow. Pattern-level corroboration is now repetitive; revisit only for an actual adopter or measured BDFL outcome.
2026-07-29T19:21:56Z
evidence attached: reddit.post.1va4wfp — A small real-world sample provides relevant context on dependency-aware planning and coordination for multi-agent coding workflows.
2026-07-28T13:30:58Z
Another parallel implementation (Dn) reinforces the plan-then-execute agentic pattern as a common convergent design, but this remains generic pattern corroboration, not BDFL adoption or reliability evidence. Engagement is flat/repetitive. The product-specific hypothesis stays open and cold pending a real BDFL user report.
2026-07-28T13:21:55Z
evidence attached: hn.story.49083082 — A separate implementation of planned, durable agent handoffs provides useful corroboration for supervised multi-agent coding workflows.
2026-07-27T05:24:44Z
The latest trigger adds no identifiable independent use of BDFL or real-project reliability result; existing evidence continues to corroborate only the broader supervisor-workflow pattern. The product-specific hypothesis remains open but should stay cold until an adopter, implementation report, or measured outcome appears.
2026-07-27T04:21:34Z
The refreshed discussion remains generic, mixed evidence for planning gates and adds no independent BDFL use or reliability result. Product-level validation is still absent, so further engagement churn should not trigger close review without a real adopter or implementation report.
2026-07-26T02:21:44Z
The trigger adds no identifiable product-level evidence: independent reports continue to validate the broader supervisor architecture, not BDFL’s actual adoption or reliability. The case remains open but repetitive and cool pending real BDFL users or project outcomes.
2026-07-26T01:23:07Z
The latest trigger adds no substantive evidence beyond previously absorbed pattern corroboration and still shows no independent BDFL use or real-project reliability results. Attention remains repetitive rather than product-validating, so the case stays cool and open.
2026-07-26T00:23:29Z
Aesop independently implements a closely comparable split-worker, versioned-state, crash-recovery harness, strengthening the architectural pattern but not validating BDFL itself. The case remains product-specific and awaits independent BDFL adoption or real-project reliability results.
2026-07-26T00:21:02Z
evidence attached: reddit.post.1v6o630 — A concrete coding harness independently explores versioned disk state, separate worker and orchestrator models, and crash recovery, materially contextualizing the broader supervisor-workflow pattern.
2026-07-25T14:25:52Z
The new production-oriented report further validates specs, tests, approval, and environment gates as a general coding-agent reliability pattern, but it does not test BDFL or its integrated recovery and coordination workflow. Product-level validation remains absent, so repeated pattern corroboration does not advance the case.
2026-07-25T14:21:17Z
evidence attached: reddit.post.1v69lvd — A firsthand coding-agent deployment describes enforcing specs, tests, and environment gates before letting agents modify sensitive production code, materially contextualizing approval-and-isolation harness patterns.
2026-07-25T03:22:41Z
The added user report further supports planning and approval gates as a general reliability pattern, but still provides no independent use or validation of BDFL’s integrated workflow. The product-level hypothesis remains open and cool pending real adopters or project results.
2026-07-25T03:20:54Z
evidence attached: reddit.post.1v5vvey — Independent user experience corroborates that upfront planning, approval gates, scoped steps, and recoverable branches improve coding-agent reliability.
2026-07-24T00:21:22Z
The latest look adds no independent BDFL adoption or reliability evidence; it only repeats support for the broader deterministic-supervisor pattern. Product-level validation remains open and the case stays cool pending real project use.
2026-07-23T21:25:05Z
Independent examples strengthen the broader deterministic supervisor pattern, but neither demonstrates use of BDFL itself or validates its planning, isolation, verification, and recovery claims. The case is now worth watching for real adopters, while remaining uncorroborated at the product level.
2026-07-23T21:21:19Z
evidence attached: reddit.post.1v4r151 — An independent implementation reports a supervisor queue, unattended workers, decision routing, and validation failures, directly corroborating the case's harness pattern.
2026-07-23T21:21:19Z
evidence attached: reddit.post.1v4r3p4 — The schema-gated execution proposal independently contextualizes the case for moving approval and execution authority into deterministic harness layers.
2026-07-23T18:25:35Z
grounded: known/medium — The architecture is already held in Scott’s Markdown OS, Long-Running Agents, and Deterministic Agent Control Plane pages, and is implemented in Proposal Compil
2026-07-23T18:22:58Z
origin walked (codex/luna, conf 0.96): anchor hn.story.49025198 -> echo.github.0e2d59b259 by Nishant Hada
2026-07-23T18:21:49Z
case created — BDFL is a concrete open-source supervisor whose integrated planning, isolation, verification, and recovery design is distinct from existing cost-focused model-orchestration cases.

Decision trace

08-03 08:22expireRepeated parallel implementations now establish the broader supervisor-workflow pattern, but sustained monitoring has produced no independent BDFL adopter, feature verification, or real-project outcom
08-03 08:20mark_dirtyengagement_update
08-03 07:21repriceThe five-month o8 deployment is the strongest practical validation yet of the broader supervisor pattern, including isolated worktrees, shared coordination, and human-approved integration. It still do
08-03 07:21attachIndependent open-source use of an orchestrator, isolated worktrees, shared coordination, and human-approved merges directly tests the supervisor workflow hypothesis.
08-03 07:20propose_attachIndependent open-source use of an orchestrator, isolated worktrees, shared coordination, and human-approved merges directly tests the supervisor workflow hypothesis.
08-02 23:22repriceThe await-human primitive adds another independent implementation of the broader approval-gate pattern, but provides no evidence of BDFL adoption, feature reliability, or real-project outcomes. Produc
08-02 23:21attachThe await-human primitive materially contextualizes the open question of whether explicit human approval and recovery workflows make multi-agent coding harnesses reliable.
08-02 23:21propose_attachThe await-human primitive materially contextualizes the open question of whether explicit human approval and recovery workflows make multi-agent coding harnesses reliable.
08-02 03:25repriceThe Codex–Claude loop is another weak parallel implementation of cross-agent orchestration, but it supplies no independent BDFL adoption, feature verification, or measured reliability result. Pattern
08-02 03:21attachIndependent implementation of a Codex-and-Claude-Code autonomous loop materially bears on whether cross-agent coding orchestration works in practice.
08-02 03:21propose_attachIndependent implementation of a Codex-and-Claude-Code autonomous loop materially bears on whether cross-agent coding orchestration works in practice.
07-30 05:25repriceThe dependency-aware WBS report adds a small practical example of scoped, checked multi-agent coordination, but still does not demonstrate independent BDFL use or validate its integrated recovery work
07-30 05:21attachA small real-world sample provides relevant context on dependency-aware planning and coordination for multi-agent coding workflows.
07-30 05:20propose_attachA small real-world sample provides relevant context on dependency-aware planning and coordination for multi-agent coding workflows.
07-28 23:30repriceAnother parallel implementation (Dn) reinforces the plan-then-execute agentic pattern as a common convergent design, but this remains generic pattern corroboration, not BDFL adoption or reliability ev
07-28 23:21attachA separate implementation of planned, durable agent handoffs provides useful corroboration for supervised multi-agent coding workflows.
07-28 23:21propose_attachA separate implementation of planned, durable agent handoffs provides useful corroboration for supervised multi-agent coding workflows.
07-27 15:24repriceThe latest trigger adds no identifiable independent use of BDFL or real-project reliability result; existing evidence continues to corroborate only the broader supervisor-workflow pattern. The product
07-27 15:20mark_dirtycomment_update
07-27 14:21repriceThe refreshed discussion remains generic, mixed evidence for planning gates and adds no independent BDFL use or reliability result. Product-level validation is still absent, so further engagement chur
07-26 12:21repriceThe trigger adds no identifiable product-level evidence: independent reports continue to validate the broader supervisor architecture, not BDFL’s actual adoption or reliability. The case remains open
07-26 12:20mark_dirtyengagement_update
07-26 11:23repriceThe latest trigger adds no substantive evidence beyond previously absorbed pattern corroboration and still shows no independent BDFL use or real-project reliability results. Attention remains repetiti
07-26 11:20mark_dirtyengagement_update
07-26 10:23repriceAesop independently implements a closely comparable split-worker, versioned-state, crash-recovery harness, strengthening the architectural pattern but not validating BDFL itself. The case remains prod
07-26 10:21attachA concrete coding harness independently explores versioned disk state, separate worker and orchestrator models, and crash recovery, materially contextualizing the broader supervisor-workflow pattern.
07-26 10:20propose_attachA concrete coding harness independently explores versioned disk state, separate worker and orchestrator models, and crash recovery, materially contextualizing the broader supervisor-workflow pattern.
07-26 00:25repriceThe new production-oriented report further validates specs, tests, approval, and environment gates as a general coding-agent reliability pattern, but it does not test BDFL or its integrated recovery a
07-26 00:21attachA firsthand coding-agent deployment describes enforcing specs, tests, and environment gates before letting agents modify sensitive production code, materially contextualizing approval-and-isolation ha
07-26 00:20propose_attachA firsthand coding-agent deployment describes enforcing specs, tests, and environment gates before letting agents modify sensitive production code, materially contextualizing approval-and-isolation ha
07-25 13:22repriceThe added user report further supports planning and approval gates as a general reliability pattern, but still provides no independent use or validation of BDFL’s integrated workflow. The product-leve
07-25 13:20attachIndependent user experience corroborates that upfront planning, approval gates, scoped steps, and recoverable branches improve coding-agent reliability.
07-25 13:20propose_attachIndependent user experience corroborates that upfront planning, approval gates, scoped steps, and recoverable branches improve coding-agent reliability.
07-24 10:21repriceThe latest look adds no independent BDFL adoption or reliability evidence; it only repeats support for the broader deterministic-supervisor pattern. Product-level validation remains open and the case
07-24 10:20mark_dirtyengagement_update
07-24 07:25repriceIndependent examples strengthen the broader deterministic supervisor pattern, but neither demonstrates use of BDFL itself or validates its planning, isolation, verification, and recovery claims. The c
07-24 07:21attachAn independent implementation reports a supervisor queue, unattended workers, decision routing, and validation failures, directly corroborating the case's harness pattern.
07-24 07:21attachThe schema-gated execution proposal independently contextualizes the case for moving approval and execution authority into deterministic harness layers.
07-24 07:20propose_attachAn independent implementation reports a supervisor queue, unattended workers, decision routing, and validation failures, directly corroborating the case's harness pattern.
07-24 07:20propose_attachThe schema-gated execution proposal independently contextualizes the case for moving approval and execution authority into deterministic harness layers.