Gholinejad, Mohammad Mehdi · Oct 22, 2026
arXiv cs.AI: Superintelligent Software Engineering: Governance That Scales with Capability
Markdown · JSON · Original source
Mohammad Mehdi Gholinejad argues that software agents more capable than human engineers will not, by capability alone, produce better software engineering. His central point is that important project decisions and contracts sit outside the code, while project knowledge decays. When agents outpace reviewers, reading catches a shrinking share of changes built on stale knowledge. A simple model suggests that, beyond a fixed review capacity, escaping changes grow with throughput while the share caught by reading approaches zero; governance can lower that growth, the author argues, whereas capability alone cannot. An exploratory snapshot of popular public GitHub repositories found agent context files in some repositories, and unresolved repository paths in many files that named paths. Gholinejad presents OSTEON, a discipline in which projects maintain a governed knowledge repository of decisions and contracts, mechanically check each change against it, check knowledge for decay at session start, gate planning, and record what sessions learn without writing rules. The paper reports no measured effect for OSTEON. It marks claims about its mechanisms as opinion and practice heuristics as design rationale, and identifies questions that could falsify its thesis.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
cs.AI · Article
Source details
Original source date: Oct 22, 2026 (2026-10-22).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-10T10:47:58.913Z. First observed: 2026-10-23T05:57:05.627Z. Retrieved: 2026-11-03T05:47:07.253Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The source describes the repository snapshot as exploratory; its findings concern the sampled repositories and do not establish how common these files are across software projects generally. The paper reports no measured effect of OSTEON, so the evidence here does not establish whether the proposed discipline improves software engineering outcomes. The supplied evidence is the paper's abstract and publisher-page material, not the full paper; details of the model and the questions proposed to falsify the thesis are not included.
Open question
What evidence would falsify the author's thesis that governance can constrain changes escaping review as agent throughput rises, while capability alone cannot?
Source-stated
Gholinejad argues that greater software-agent capability alone does not produce better software engineering, because correctness depends on decisions and contracts not carried in code, and knowledge decays.
capability alone does not produce better software engineering
Source-stated
The paper's simple model says that past a fixed review capacity, escaping changes grow linearly with throughput while the share caught by reading falls toward zero; the author says governance can lower the slope.
past a fixed review capacity, escaping changes grow linearly with throughput while the share that reading catches falls toward zero; governance can lower the slope, capability alone cannot.
Source-stated
An exploratory snapshot of 452 of the most-starred public GitHub repositories found a root-level or .github agent context file in 48.67%; among median-file paths, 9.5% to 22% did not resolve, depending on how a reference is read.
An exploratory snapshot of 452 of the most-starred public GitHub repositories finds a root-level or .github agent context file in 48.67%
Source-stated
OSTEON is presented as a discipline using a governed knowledge repository, mechanical checks against it, decay checks at session start, gated planning, and session records that do not write rules.
every change is checked mechanically against it, and the knowledge itself is checked for decay at every session start; planning is gated
Source-stated
The paper reports no measured effect of OSTEON and labels mechanism claims as opinion and practice heuristics as design rationale.
We report no measured effect of OSTEON, mark claims about its mechanisms as opinion and the author's practice heuristics as design rationale
Source and contributor context
Gholinejad, Mohammad Mehdi · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Superintelligence
A hypothetical AI far more capable than people across most intellectual tasks.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: f06f8055-2b16-4793-847b-b16568507315.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-10T10:47:58.913Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: 1a1060d0596a9633de2783f10059b39868f63939dc0925ba39ec3d5cee9517e4. Source text hash: b3f53380747758fecc0b69ec767ae3a634d7be8820bf8a19a4bb49802bd63623.
Revision and source-check history
Version 1 · 2026-11-10T10:47:58.913Z
arXiv cs.AI: Superintelligent Software Engineering: Governance That Scales with Capability
Mohammad Mehdi Gholinejad argues that software agents more capable than human engineers will not, by capability alone, produce better software engineering. His central point is that important project decisions and contracts sit outside the code, while project knowledge decays. When agents outpace reviewers, reading catches a shrinking share of changes built on stale knowledge. A simple model suggests that, beyond a fixed review capacity, escaping changes grow with throughput while the share caught by reading approaches zero; governance can lower that growth, the author argues, whereas capability alone cannot. An exploratory snapshot of popular public GitHub repositories found agent context files in some repositories, and unresolved repository paths in many files that named paths. Gholinejad presents OSTEON, a discipline in which projects maintain a governed knowledge repository of decisions and contracts, mechanically check each change against it, check knowledge for decay at session start, gate planning, and record what sessions learn without writing rules. The paper reports no measured effect for OSTEON. It marks claims about its mechanisms as opinion and practice heuristics as design rationale, and identifies questions that could falsify its thesis.
Original source
cs.AI · Article
Source details
Original source date: Oct 22, 2026 (2026-10-22).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-10T10:47:58.913Z. First observed: 2026-10-23T05:57:05.627Z. Retrieved: 2026-11-03T05:47:07.253Z.
Limits
The source describes the repository snapshot as exploratory; its findings concern the sampled repositories and do not establish how common these files are across software projects generally. The paper reports no measured effect of OSTEON, so the evidence here does not establish whether the proposed discipline improves software engineering outcomes. The supplied evidence is the paper's abstract and publisher-page material, not the full paper; details of the model and the questions proposed to falsify the thesis are not included.
Open question
What evidence would falsify the author's thesis that governance can constrain changes escaping review as agent throughput rises, while capability alone cannot?
Source-stated
Gholinejad argues that greater software-agent capability alone does not produce better software engineering, because correctness depends on decisions and contracts not carried in code, and knowledge decays.
capability alone does not produce better software engineering
Source-stated
The paper's simple model says that past a fixed review capacity, escaping changes grow linearly with throughput while the share caught by reading falls toward zero; the author says governance can lower the slope.
past a fixed review capacity, escaping changes grow linearly with throughput while the share that reading catches falls toward zero; governance can lower the slope, capability alone cannot.
Source-stated
An exploratory snapshot of 452 of the most-starred public GitHub repositories found a root-level or .github agent context file in 48.67%; among median-file paths, 9.5% to 22% did not resolve, depending on how a reference is read.
An exploratory snapshot of 452 of the most-starred public GitHub repositories finds a root-level or .github agent context file in 48.67%
Source-stated
OSTEON is presented as a discipline using a governed knowledge repository, mechanical checks against it, decay checks at session start, gated planning, and session records that do not write rules.
every change is checked mechanically against it, and the knowledge itself is checked for decay at every session start; planning is gated
Source-stated
The paper reports no measured effect of OSTEON and labels mechanism claims as opinion and practice heuristics as design rationale.
We report no measured effect of OSTEON, mark claims about its mechanisms as opinion and the author's practice heuristics as design rationale
Source and contributor context
Gholinejad, Mohammad Mehdi · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Superintelligence
A hypothetical AI far more capable than people across most intellectual tasks.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: f06f8055-2b16-4793-847b-b16568507315.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-10T10:47:58.913Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: 1a1060d0596a9633de2783f10059b39868f63939dc0925ba39ec3d5cee9517e4. Source text hash: b3f53380747758fecc0b69ec767ae3a634d7be8820bf8a19a4bb49802bd63623.
uncertain · 2026-11-10T10:47:58.913Z · anthropic/claude-opus-5.5
Publication: f06f8055-2b16-4793-847b-b16568507315
attribution — pass: The reading attributes the argument, model, snapshot and OSTEON to the sole listed author, Mohammad Mehdi Gholinejad. It keeps the thesis as his argument rather than established fact, including "the author argues."
Authors:Mohammad Mehdi Gholinejad
summary — uncertain: Mostly faithful, but two passages weaken meaning. First, 'found agent context files in some repositories' erases the headline figure: nearly half (48.67%) of 452 repositories. Repair: state the figure. Second, central terms are unexplained for a general reader, including 'agent context file', 'contracts', 'escaping changes' and 'governance'. A tooltip or short gloss is needed. The summary also drops 'linearly' from the model claim and omits that OSTEON is a Claude Code plugin.
finds a root-level or .github agent context file in 48.67% (95% confidence interval 44.1 to 53.3)
date — pass: originalPublished is 2026-10-22, which matches the v1 submission date. The v2 revision on 29 Oct and the feed date of 2 Nov are correctly not used.
Submitted on 22 Oct 2026 (v1), last revised 29 Oct 2026
support — pass: Every claim matches the abstract. The reading preserves the conditional model claim, the exploratory scope of the snapshot, and the stated absence of any measured OSTEON effect. The limitations are accurate. The question follows from the author's own statement that falsifying questions exist, without asserting an answer.
We report no measured effect of OSTEON, mark claims about its mechanisms as opinion
media — pass: The selected media is the verified direct arXiv abstract page for this paper. It is the most direct access to the source.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXiv