diff --git a/site/ai-agent-no-progress/index.html b/site/ai-agent-no-progress/index.html new file mode 100644 index 0000000..73c2218 --- /dev/null +++ b/site/ai-agent-no-progress/index.html @@ -0,0 +1,69 @@ + + + + + +AI Agent No-Progress Loops — Detection & Governance + + + + + + + + + + + + + + + + + + + + + + +
+
+
MARGINAL / Guides / No-progress loops
+
+

No-progress loops

+

AI agents repeat work.Progress is what matters.

+

A practical model for detecting successful agent activity that repeats without changing observable state or producing new evidence.

+
Watch the demo ↗View on GitHub ↗
+ +
+ +

The signal

Activity is not progress.

+

A coding agent can execute a tool successfully and still learn nothing new. The useful question is not “did the call succeed?” but did observable state or useful evidence change?

+
same semantic action ++ successful outcome ++ unchanged observable workspace state ++ no new evidence +→ no-progress repetition candidate
+

MARGINAL treats that pattern as evidence of diminishing marginal value. It does not label every retry as waste: failures, changed state, new evidence and unsupported outcomes remove or reset repetition pressure.

+

Why fixed token caps are not enough

A budget cap can stop a run after a threshold, but it cannot tell whether the next verification call is useful. No-progress governance is state-aware: the decision depends on what changed, what was learned and what the runtime can actually prove.

+
+

Mechanism

Observe → prove → earn.

+
Observe

Collect action identity, outcomes, state and derived evidence locally. Raw prompts and source are not required as evidence fields.

+
Prove

Look for repeated successful work with unchanged state and no new evidence. Ambiguity fails open.

+
Earn

Only an adapter with real blocking capability and sufficient local evidence can move beyond advisory behavior.

+
+

No universal savings claim

MARGINAL measures governance overhead and workload outcomes. A historical exploratory smoke observed a token difference but did not establish causal savings; the project keeps that distinction explicit.

+ +

FAQ

Fast answers.

What is a no-progress loop in an AI agent?

Repeated work where the semantic action succeeds but observable workspace state and useful evidence do not change.

Does MARGINAL stop every repeated action?

No. Verification, failures, changed state, new evidence and ambiguous outcomes are treated conservatively. Unsupported or ambiguous cases fail open.

Is MARGINAL just a token limiter?

No. Token budgets are one cost signal; MARGINAL focuses on whether another action is expected to create enough verified value to justify its cost, latency and risk.

+

Try it

Observe first. Prove waste. Earn enforcement.

+

Install MARGINAL in Shadow Mode, inspect what your agent actually repeats, and contribute traces that make the governor harder to fool.

+
★ Star MARGINAL on GitHubExplore more guides
+
+
+ + diff --git a/site/claude-code/index.html b/site/claude-code/index.html new file mode 100644 index 0000000..d41fbfd --- /dev/null +++ b/site/claude-code/index.html @@ -0,0 +1,65 @@ + + + + + +MARGINAL for Claude Code — Observe No-Progress Loops + + + + + + + + + + + + + + + + + + + + + + +
+
+
MARGINAL / Guides / Claude Code
+
+

Claude Code

+

See no-progress loops.Do not pretend to control them.

+

MARGINAL integrates with Claude Code in Observe-only mode: collect evidence, detect repetition and keep the execution path unchanged.

+
Watch the demo ↗View on GitHub ↗
+ +
+ +

Claude Code integration

Observe repeated work without changing Claude Code behavior.

+

MARGINAL's Claude Code integration is intentionally labeled Observe. Native hooks normalize lifecycle evidence into the same provider-neutral governance model, but the adapter declares no blocking capability and does not deny tool calls.

+
marginal install claude-code
+

Claude Code exposes separate successful and failed post-tool events, so MARGINAL can record the outcome as an engine-declared fact instead of inferring it from response text. Per-tool token usage is not exposed through that hook surface and is reported as unavailable rather than zero.

+

Why Observe-only is a feature, not a missing disclaimer

An integration should not claim authority it does not possess. MARGINAL can detect repeated-work recommendations and governance overhead while preserving Claude Code's next action. That evidence can inform future work without pretending enforcement already exists.

+
+

What it records

Structured evidence, conservative outcomes.

+
Engine-declared

Success / failure

Separate hook events provide a stronger outcome signal than guessing from free text.

+
Local

Decision Ledger

Recommendations and evidence are recorded without turning the hook into an enforcement surface.

+
Unavailable ≠ zero

Token usage

If the hook surface does not expose per-tool tokens, MARGINAL reports that honestly.

+
Observe

No blocking

The adapter declares no control capability, so repeated-work candidates remain recommendations.

+
+ +

FAQ

Fast answers.

Does MARGINAL block Claude Code tool calls?

No. The current Claude Code adapter is Observe-only and declares no blocking capability.

Can it still detect no-progress repetition?

Yes. It can normalize action, outcome and workspace-state evidence and record repeated-work recommendations in the Decision Ledger.

Does it read Claude Code prompts or transcripts for evidence?

The integration is designed around structured derived signals rather than persisting raw prompts, command text, source paths or transcripts as governance evidence.

+

Try it

Observe first. Prove waste. Earn enforcement.

+

Install MARGINAL in Shadow Mode, inspect what your agent actually repeats, and contribute traces that make the governor harder to fool.

+
★ Star MARGINAL on GitHubExplore more guides
+
+
+ + diff --git a/site/codex/index.html b/site/codex/index.html new file mode 100644 index 0000000..05f9b79 --- /dev/null +++ b/site/codex/index.html @@ -0,0 +1,66 @@ + + + + + +MARGINAL for Codex — No-Progress Runtime Governance + + + + + + + + + + + + + + + + + + + + + + +
+
+
MARGINAL / Guides / Codex
+
+

Codex

+

Runtime governance for Codex.Shadow first.

+

A native Codex plugin that observes repeated work, records evidence locally and only earns narrow enforcement after proof.

+
Watch the demo ↗View on GitHub ↗
+ +
+ +

Codex integration

Install once. Start powerless.

+

MARGINAL ships a native Codex plugin. A new installation begins in Shadow Mode: it observes lifecycle signals and records governance evidence without assuming that installation grants authority to block.

+
codex plugin marketplace add SignalLayerLabs/Marginal --ref main +codex plugin add marginal@marginal
+

The Codex adapter is currently labeled Tool Enforcement, not Full Compute Enforcement. Its authority is deliberately narrow and repository-local: only supported tool actions can become denyable after representative evidence and explicit promotion.

+

What Earned Enforcement means

Promotion is evidence-bound rather than global. Integrity drift, changed evidence or safety failures can demote the repository back toward observation. Generic shell, tests, search, writes, network and unsupported paths remain outside broad enforcement claims.

+
+

Trust boundary

Capability claims stay adapter-specific.

+
Shadow

Observe actions and decisions without changing caller behavior.

+
Tool Enforcement

Supported Codex tool actions can be intercepted under the exact proven condition.

+
Not claimed

MARGINAL does not describe Codex integration as Full Compute Enforcement.

+
+

Best first test

Install the plugin, use Codex normally, then inspect MARGINAL status/diagnostics and the Decision Ledger. The first useful outcome is evidence about your workload—not a universal savings percentage.

+ +

FAQ

Fast answers.

Does MARGINAL block Codex immediately after install?

No. It starts in Shadow Mode. Narrow repository-local Tool Enforcement must be earned from local evidence and explicit promotion.

Does MARGINAL control every Codex action?

No. The public capability is Tool Enforcement, not Full Compute Enforcement, and unsupported or ambiguous paths remain outside the blocking boundary.

Can I remove the Codex plugin?

Yes. The project documents a clean removal path using codex plugin remove marginal@marginal.

+

Try it

Observe first. Prove waste. Earn enforcement.

+

Install MARGINAL in Shadow Mode, inspect what your agent actually repeats, and contribute traces that make the governor harder to fool.

+
★ Star MARGINAL on GitHubExplore more guides
+
+
+ + diff --git a/site/guides/index.html b/site/guides/index.html new file mode 100644 index 0000000..700133e --- /dev/null +++ b/site/guides/index.html @@ -0,0 +1,16 @@ + +MARGINAL Guides — AI Agent No-Progress Governance + + + + +
+

Technical guides

Search the problem.Inspect the mechanism.

Evergreen, evidence-first guides for developers who see coding agents repeat reads, searches, checks or tool calls without observable progress.

+
+AI agent no-progress loopsWhat no-progress means, how to detect it, and why retries are not automatically waste. +Repeated tool callsDistinguish useful verification from same-state repetition with no new evidence. +MARGINAL for CodexNative plugin, Shadow Mode by default, narrow evidence-earned Tool Enforcement. +MARGINAL for Claude CodeObserve-only hooks, engine-declared outcomes, local Decision Ledger, no blocking. +
+

Bring us an agent that gets stuck.

Real traces and falsifying examples are more useful than generic feature requests.

Contribute an example ↗Watch the demo
+
\ No newline at end of file diff --git a/site/index.html b/site/index.html index 020b50b..29658c3 100644 --- a/site/index.html +++ b/site/index.html @@ -1 +1 @@ -MARGINAL — Stop No-Progress Loops in AI Coding Agents

Runtime governor for AI coding agents

AI agents repeat work that changed nothing.MARGINAL catches it.

MARGINAL watches tool actions, outcomes and workspace evidence. When the same successful action repeats with no observable progress, it can identify the loop — without assuming every retry is waste.

Observe first. Prove waste. Earn enforcement.

★ Star MARGINAL on GitHubWatch the mechanism ↓
  • Open source
  • Local first
  • Provider neutral
  • Zero mandatory runtime dependencies
agent trace / workspace
01
Read config.pynew evidence acquired
RUN
02
Read config.pyverification; outcome successful
RUN
03
Read config.pysame workspace state · no new evidence
OBSERVE
04
Read config.pyexact eligible no-progress repetition
STOP*
*Only after local Earned Enforcement. Otherwise MARGINAL stays advisory.Fails open on ambiguity.

10-second mechanism demo

Activity is not the same thing as progress.

Same action. Same state. No new evidence. That is the signal MARGINAL cares about. This deterministic visual is a mechanism demonstration, not a production benchmark.

Without a governorLOOP
01Read config.pyRUN
02Read config.pyRUN
03Read config.pyRUN
04Read config.pyRUN
05Read config.pyRUN
With MARGINALEVIDENCE
01Read config.pyNEW EVIDENCE
02Read config.pyVERIFY
03Read config.pySAME STATE
04Read config.pySTOP CANDIDATE
05Earned authority?BLOCK / ALLOW
Open shareable demo ↗No API credits required. No provider telemetry claimed.

How it works

A governor that has to earn the right to govern.

Installation does not equal authority. MARGINAL separates observation, proof and enforcement so an efficiency tool cannot casually become a correctness risk.

01 / OBSERVE

Watch

Collect derived action, outcome, coverage, workspace-state and evidence signals locally.

02 / PROVE

Compare

Look for repeated successful actions where observable state and useful evidence did not change.

03 / EARN

Build trust

Require representative local evidence, clean coverage and explicit promotion before blocking.

04 / INTERVENE

Stop narrowly

Only eligible action families can be denied, and only under the exact proven no-progress condition.

05 / RECOVER

Fail open

Unknown outcomes, drift, integrity failures or changed evidence remove pressure and restore allowance.

Works alongside coding agents

One governance core. Conservative engine boundaries.

MARGINAL is not another coding agent. It sits beside supported runtimes and turns native lifecycle signals into the same provider-neutral evidence model.

Tool Enforcement

Codex

Native plugin. Shadow Mode by default; repository-local Earned Enforcement is intentionally narrow.

Observe-only

Claude Code

Native hooks record engine-declared success and failure without changing the next model action.

Observe-only

OpenCode

In-process JavaScript plugin with a persistent stdio bridge to the provider-neutral runtime.

Observe-only

PrivacyCode

OpenCode-compatible target with its own engine identity, ledger root and trust evidence.

Designed to be falsifiable

MARGINAL has to justify its own overhead, too.

The project treats governance cost, harmful interventions and preserved quality as first-class measurements — not footnotes.

01

Shadow first

New installations observe before blocking. Lack of evidence is not permission.

02

Decision ledger

Canonical records are hash chained so decisions can be reproduced and integrity drift detected.

03

Local-first privacy

Raw prompts, source, commands, outputs, transcripts and credentials are not evidence fields.

04

Fail open

Ambiguous or unsupported outcomes do not become evidence for blocking.

05

Explicit boundaries

Tool Enforcement is not presented as Full Compute Enforcement. Capabilities stay adapter-specific.

06

Graceful irrelevance

If a future agent is already efficient, MARGINAL should measure that and get out of the way.

Public evidence, without marketing math

The first smoke validated the integration — not the savings claim.

We keep negative and inconclusive results public because MARGINAL's credibility depends on separating observation from causation.

Exploratory SWE-bench Lite smoke

Exploratory 3-task smoke, one paired run per task. Both lanes resolved 0/3. No deny was applied in these three agent trajectories. A 24.93% token difference was observed, so the difference is not attributed to MARGINAL and is not a support claim.

The full report, raw JSON, protocol and evidence bundle remain public in the repository.

Read public report ↗Inspect raw evidence ↗
Resolved0/3 → 0/3
Effective tokens24.93% fewer
Tool calls3.03% fewer
Governance latency7.06 s
Applied denies0
Evaluatorpass_through

Install

Start in Shadow Mode.

The Codex marketplace install is one command. Enforcement still has to be earned locally.

codex plugin marketplace add SignalLayerLabs/Marginal --ref main && codex plugin add marginal@marginal

Remove cleanly with codex plugin remove marginal@marginal.

FAQ

Fast answers before you clone.

What does MARGINAL actually stop?

Today, Codex Tool Enforcement is deliberately narrow. Exact eligible workspace-local reads can become denyable after verified repeated success with no state or evidence change. Generic shell, tests, search, writes, network and unknown MCP paths remain observe/recommend only.

Is MARGINAL a security product?

No. It is an efficiency governor, not a security boundary against software running as the same OS user.

Does it work with Claude Code?

Yes in Observe-only mode. Claude Code hooks can record engine-declared success/failure and feed the same evidence model, but the integration does not block actions today.

Why not just cap tokens?

A fixed cap cannot distinguish useful verification from repeated no-progress work. MARGINAL focuses on the marginal value of the next action and on observable evidence, not only the size of a budget.

Open source · Apache-2.0

If your coding agent loops, make the loop prove it is useful.

Clone it, inspect the hooks, run Shadow Mode, challenge the evidence model — and star the repo if you want this idea to keep moving.

★ Star MARGINALQuickstart ↗
\ No newline at end of file +MARGINAL — Stop No-Progress Loops in AI Coding Agents

Runtime governor for AI coding agents

AI agents repeat work that changed nothing.MARGINAL catches it.

MARGINAL watches tool actions, outcomes and workspace evidence. When the same successful action repeats with no observable progress, it can identify the loop — without assuming every retry is waste.

Observe first. Prove waste. Earn enforcement.

★ Star MARGINAL on GitHubWatch the mechanism ↓
  • Open source
  • Local first
  • Provider neutral
  • Zero mandatory runtime dependencies
agent trace / workspace
01
Read config.pynew evidence acquired
RUN
02
Read config.pyverification; outcome successful
RUN
03
Read config.pysame workspace state · no new evidence
OBSERVE
04
Read config.pyexact eligible no-progress repetition
STOP*
*Only after local Earned Enforcement. Otherwise MARGINAL stays advisory.Fails open on ambiguity.

10-second mechanism demo

Activity is not the same thing as progress.

Same action. Same state. No new evidence. That is the signal MARGINAL cares about. This deterministic visual is a mechanism demonstration, not a production benchmark.

Without a governorLOOP
01Read config.pyRUN
02Read config.pyRUN
03Read config.pyRUN
04Read config.pyRUN
05Read config.pyRUN
With MARGINALEVIDENCE
01Read config.pyNEW EVIDENCE
02Read config.pyVERIFY
03Read config.pySAME STATE
04Read config.pySTOP CANDIDATE
05Earned authority?BLOCK / ALLOW
Open shareable demo ↗No API credits required. No provider telemetry claimed.

How it works

A governor that has to earn the right to govern.

Installation does not equal authority. MARGINAL separates observation, proof and enforcement so an efficiency tool cannot casually become a correctness risk.

01 / OBSERVE

Watch

Collect derived action, outcome, coverage, workspace-state and evidence signals locally.

02 / PROVE

Compare

Look for repeated successful actions where observable state and useful evidence did not change.

03 / EARN

Build trust

Require representative local evidence, clean coverage and explicit promotion before blocking.

04 / INTERVENE

Stop narrowly

Only eligible action families can be denied, and only under the exact proven no-progress condition.

05 / RECOVER

Fail open

Unknown outcomes, drift, integrity failures or changed evidence remove pressure and restore allowance.

Works alongside coding agents

One governance core. Conservative engine boundaries.

MARGINAL is not another coding agent. It sits beside supported runtimes and turns native lifecycle signals into the same provider-neutral evidence model.

Tool Enforcement

Codex

Native plugin. Shadow Mode by default; repository-local Earned Enforcement is intentionally narrow.

Observe-only

Claude Code

Native hooks record engine-declared success and failure without changing the next model action.

Observe-only

OpenCode

In-process JavaScript plugin with a persistent stdio bridge to the provider-neutral runtime.

Observe-only

PrivacyCode

OpenCode-compatible target with its own engine identity, ledger root and trust evidence.

Designed to be falsifiable

MARGINAL has to justify its own overhead, too.

The project treats governance cost, harmful interventions and preserved quality as first-class measurements — not footnotes.

01

Shadow first

New installations observe before blocking. Lack of evidence is not permission.

02

Decision ledger

Canonical records are hash chained so decisions can be reproduced and integrity drift detected.

03

Local-first privacy

Raw prompts, source, commands, outputs, transcripts and credentials are not evidence fields.

04

Fail open

Ambiguous or unsupported outcomes do not become evidence for blocking.

05

Explicit boundaries

Tool Enforcement is not presented as Full Compute Enforcement. Capabilities stay adapter-specific.

06

Graceful irrelevance

If a future agent is already efficient, MARGINAL should measure that and get out of the way.

Public evidence, without marketing math

The first smoke validated the integration — not the savings claim.

We keep negative and inconclusive results public because MARGINAL's credibility depends on separating observation from causation.

Exploratory SWE-bench Lite smoke

Exploratory 3-task smoke, one paired run per task. Both lanes resolved 0/3. No deny was applied in these three agent trajectories. A 24.93% token difference was observed, so the difference is not attributed to MARGINAL and is not a support claim.

The full report, raw JSON, protocol and evidence bundle remain public in the repository.

Read public report ↗Inspect raw evidence ↗
Resolved0/3 → 0/3
Effective tokens24.93% fewer
Tool calls3.03% fewer
Governance latency7.06 s
Applied denies0
Evaluatorpass_through

Install

Start in Shadow Mode.

The Codex marketplace install is one command. Enforcement still has to be earned locally.

codex plugin marketplace add SignalLayerLabs/Marginal --ref main && codex plugin add marginal@marginal

Remove cleanly with codex plugin remove marginal@marginal.

Technical guides

Search the problem. Inspect the mechanism.

Evidence-first guides for developers dealing with repeated reads, tool calls and no-progress loops in coding agents.

01

No-progress loops

Detect successful activity that repeats without observable progress.

03

Codex

Shadow-first native plugin with narrow evidence-earned Tool Enforcement.

04

Claude Code

Observe-only hooks with engine-declared outcomes and no blocking.

Explore all guides →

FAQ

Fast answers before you clone.

What does MARGINAL actually stop?

Today, Codex Tool Enforcement is deliberately narrow. Exact eligible workspace-local reads can become denyable after verified repeated success with no state or evidence change. Generic shell, tests, search, writes, network and unknown MCP paths remain observe/recommend only.

Is MARGINAL a security product?

No. It is an efficiency governor, not a security boundary against software running as the same OS user.

Does it work with Claude Code?

Yes in Observe-only mode. Claude Code hooks can record engine-declared success/failure and feed the same evidence model, but the integration does not block actions today.

Why not just cap tokens?

A fixed cap cannot distinguish useful verification from repeated no-progress work. MARGINAL focuses on the marginal value of the next action and on observable evidence, not only the size of a budget.

Open source · Apache-2.0

If your coding agent loops, make the loop prove it is useful.

Clone it, inspect the hooks, run Shadow Mode, challenge the evidence model — and star the repo if you want this idea to keep moving.

★ Star MARGINALQuickstart ↗
\ No newline at end of file diff --git a/site/reach.css b/site/reach.css new file mode 100644 index 0000000..94027f6 --- /dev/null +++ b/site/reach.css @@ -0,0 +1,32 @@ +/* Reach Moat landing-page layer. Keeps the existing MARGINAL design tokens. */ +.reach-hero{padding:92px 0 68px;max-width:900px} +.reach-hero h1{margin:0;font-size:clamp(3rem,7vw,6rem);line-height:.94;letter-spacing:-.055em} +.reach-hero h1 span{display:block;color:var(--acid)} +.reach-hero .hero-lead{max-width:780px} +.reach-grid{display:grid;grid-template-columns:repeat(2,1fr);gap:14px} +.reach-card{padding:24px;border:1px solid var(--line);border-radius:16px;background:var(--panel)} +.reach-card h3{margin:12px 0 8px;font-size:1.35rem} +.reach-card p{margin:0;color:var(--muted)} +.reach-card .mode{display:inline-flex;padding:4px 8px;border:1px solid var(--line);border-radius:999px;color:var(--acid);font:800 .68rem var(--mono);letter-spacing:.08em;text-transform:uppercase} +.reach-prose{max-width:850px} +.reach-prose h2{margin:0 0 18px;font-size:clamp(2rem,4vw,3.7rem);line-height:1;letter-spacing:-.04em} +.reach-prose h3{margin:34px 0 8px;font-size:1.35rem} +.reach-prose p,.reach-prose li{color:var(--muted)} +.reach-prose strong{color:var(--text)} +.reach-code{margin:22px 0;padding:18px;border:1px solid rgba(168,255,99,.32);border-radius:14px;background:rgba(168,255,99,.04);overflow:auto} +.reach-code code{color:#dfffd0;font:500 .84rem var(--mono);white-space:pre-wrap} +.reach-compare{display:grid;grid-template-columns:repeat(3,1fr);gap:12px} +.reach-compare article{padding:20px;border:1px solid var(--line);border-radius:14px;background:var(--panel)} +.reach-compare b{display:block;color:var(--acid);font:800 .75rem var(--mono);text-transform:uppercase;letter-spacing:.08em} +.reach-compare p{color:var(--muted);margin:10px 0 0} +.reach-callout{padding:24px;border:1px solid rgba(168,255,99,.36);border-radius:16px;background:linear-gradient(145deg,rgba(168,255,99,.06),var(--panel))} +.reach-callout h3{margin:0 0 8px} +.reach-callout p{margin:0;color:var(--muted)} +.breadcrumbs{padding-top:24px;color:var(--muted);font-size:.82rem} +.breadcrumbs a{text-decoration:none} +.guide-list{display:grid;grid-template-columns:repeat(2,1fr);gap:14px} +.guide-list a{display:block;padding:24px;border:1px solid var(--line);border-radius:16px;background:var(--panel);text-decoration:none} +.guide-list a:hover{border-color:rgba(168,255,99,.4)} +.guide-list strong{display:block;font-size:1.25rem} +.guide-list span{display:block;margin-top:7px;color:var(--muted)} +@media(max-width:720px){.reach-grid,.reach-compare,.guide-list{grid-template-columns:1fr}.reach-hero{padding:64px 0 50px}} diff --git a/site/repeated-tool-calls/index.html b/site/repeated-tool-calls/index.html new file mode 100644 index 0000000..994da2c --- /dev/null +++ b/site/repeated-tool-calls/index.html @@ -0,0 +1,71 @@ + + + + + +Repeated AI Agent Tool Calls — Detect Useful vs Waste + + + + + + + + + + + + + + + + + + + + + + +
+
+
MARGINAL / Guides / Repeated tool calls
+
+

Repeated tool calls

+

Same tool call.Different value.

+

Counting retries is easy. Deciding whether a retry produced progress requires state, outcomes and evidence.

+
Watch the demo ↗View on GitHub ↗
+ +
+ +

Repeated tool calls

The second read may be verification. The fourth identical read may not be.

+

Repeated tool calls are not inherently pathological. An agent may legitimately re-read a file after a write, rerun a test after changing code, or repeat a search because the environment changed. A useful detector therefore needs more than call counts.

+

Three questions before calling a repeat waste

+
  • Was the previous outcome proven? Unknown outcomes should not be promoted into evidence for blocking.
    • +
  • Did observable workspace state change? A changed repository state can make the same action valuable again.
    • +
  • Did the action add evidence? New evidence resets repetition pressure even when the tool name looks identical.
+
read config.py → new evidence +read config.py → verification +read config.py → same state, no new evidence +read config.py → no-progress candidate
+

MARGINAL turns this into a provider-neutral evidence model. The model can stay purely observational, or—only where an adapter has real control and local evidence supports promotion—participate in narrow Tool Enforcement.

+
+

Design rule

Useful retries must survive.

+
Allow pressure reset

Changed state

A write, checkout or other observable state change means an old repetition proof may no longer apply.

+
Fail open

Unknown outcome

If the runtime cannot prove success or failure, MARGINAL does not manufacture certainty.

+
Preserve correctness

Verification

Repeated work can be rational when it acquires or confirms evidence needed to complete the task safely.

+
Candidate

Same action, same state

Successful work with no observable progress is the narrow pattern worth escalating.

+
+ +

FAQ

Fast answers.

Why do AI coding agents repeat tool calls?

Retries can come from verification, uncertainty, changing state, tool failures or genuine no-progress loops. The same surface behavior can have different causes.

Can repeated reads be useful?

Yes. A read after a write or a read that yields new evidence can be useful. MARGINAL is designed not to treat repetition alone as proof of waste.

What does MARGINAL persist?

Its privacy model uses derived structured evidence; raw prompts, source, commands, outputs, transcripts and credentials are not default evidence fields.

+

Try it

Observe first. Prove waste. Earn enforcement.

+

Install MARGINAL in Shadow Mode, inspect what your agent actually repeats, and contribute traces that make the governor harder to fool.

+
★ Star MARGINAL on GitHubExplore more guides
+
+
+ + diff --git a/site/sitemap.xml b/site/sitemap.xml index 903bfa2..0ffa8e3 100644 --- a/site/sitemap.xml +++ b/site/sitemap.xml @@ -1,8 +1,8 @@ - + https://signallayerlabs.github.io/Marginal/weekly1.0 https://signallayerlabs.github.io/Marginal/demo/monthly0.8 https://signallayerlabs.github.io/Marginal/privacy.htmlmonthly0.5 https://signallayerlabs.github.io/Marginal/support.htmlmonthly0.5 https://signallayerlabs.github.io/Marginal/terms.htmlmonthly0.4 - +https://signallayerlabs.github.io/Marginal/guides/weekly0.9https://signallayerlabs.github.io/Marginal/ai-agent-no-progress/weekly0.9https://signallayerlabs.github.io/Marginal/repeated-tool-calls/weekly0.9https://signallayerlabs.github.io/Marginal/codex/weekly0.9https://signallayerlabs.github.io/Marginal/claude-code/weekly0.9 \ No newline at end of file