GrantTapGrantTap Journal
Agent work you can understand and control.
Field guides for running coding agents with confidence: see progress, make better decisions, and keep one Task moving across tools and devices. Explore comparisons, Mesh, governance, and verifiable results.
Practical agent workflows · 2026-10-13 · 15 minLight, Dark and Auto: a website theme that keeps its promiseGrantTap’s website as a practical case study: system changes, explicit overrides, storage failures, first paint, cross-tab updates and readable technical pages.Read the story ↗All stories
55 stories
GrantTap field guide · 2026-11-08 · 14 minChoose an agent that is ready for this TaskLook beyond a model name: current availability, the intended computer, authentication, tools, control coverage, privacy and a small representative result.Read story ↗
GrantTap field guide · 2026-11-07 · 13 minFrom an agent's Done to a release you can verifyConnect the user scenario, source revision, meaningful checks, artifact identity and active release. A practical evidence chain for updates you can actually find on your device.Read story ↗
GrantTap field guide · 2026-11-06 · 13 minParallel agents need separate work and one integration pathUse independent outcomes, separate worktrees, explicit shared-resource ownership and combined verification. More processes help only when the verified result arrives sooner.Read story ↗
GrantTap field guide · 2026-11-05 · 12 minContext an agent can use: fresh evidence beyond a long chatSelect context for the next decision, bind claims to revisions, keep sources expandable and preserve constraints through compaction. Measure completeness before celebrating fewer tokens.Read story ↗
GrantTap field guide · 2026-11-04 · 12 minDevices that are ready: runtime versions, login and cacheA computer can be online while a provider is unavailable. Inspect installed and running versions, authentication, owned processes and clearable storage before choosing a repair.Read story ↗
GrantTap guide · 2026-11-03 · 12 minFrom Send to Started: understand an agent message's journeyA visible message, a relay receipt and agent activity establish different facts. A practical GrantTap guide to delayed requests, fresh Working status, safe retries and meaningful latency measurements.Read story ↗
GrantTap field guide · 2026-11-02 · 13 minUseful agent automations need a real stop controlDesign recurring work around one question, an explicit destination and authority, a timezone, an overlap policy, useful results and a verified end to future runs.Read story ↗
GrantTap field guide · 2026-11-01 · 13 minWhich device can read this? A privacy map for agent workTrace source, model context, decisions and diagnostics across the computer, phone, browser, provider and relay. Encryption and endpoint trust answer different questions.Read story ↗
GrantTap field guide · 2026-10-31 · 13 minAn approval has a lifecycle beyond the Allow buttonFollow a decision from the phone to the execution boundary: request identity, stale answers, effective policy, host acknowledgment and verified continuation.Read story ↗
GrantTap field guide · 2026-10-30 · 12 minA better task brief before your coding agent startsTurn a vague request into an observable result: the failing moment, environment, authority, useful evidence and a completion scenario you can review from your phone.Read story ↗
GrantTap field guide · 2026-10-29 · 15 minBring a colleague into an agent project with clear accessA collaboration lifecycle across person, device, project and computer: scoped roles, repository grants, useful context, secrets, verified handoff and honest offboarding.Read story ↗
GrantTap field guide · 2026-10-28 · 15 minWhy is the agent computer busy? Read processes, CPU, memory and diskInvestigate local load by ownership, interval and useful work. Understand process groups, measured versus attributed use, safe cache cleanup, runtime versions and targeted restarts.Read story ↗
GrantTap field guide · 2026-10-27 · 15 minReview an agent patch from your phone without losing the evidenceA focused mobile review: identify the Task and Execution, choose the right request, read Modified and All files, inspect tests and distinguish accepted code from a published release.Read story ↗
GrantTap field guide · 2026-10-26 · 15 minWhen a document tries to instruct your coding agentSeparate authorized instructions from retrieved material. Inspect tool inputs, preserve provenance, limit credentials and verify the execution boundary without claiming a universal prompt-injection shield.Read story ↗
GrantTap field guide · 2026-10-25 · 15 minEvaluate a coding agent on work you can actually verifyA practical evaluation method for your own repository: comparable starting conditions, meaningful acceptance, human intervention, observable evidence and a specific routing decision.Read story ↗
GrantTap field guide · 2026-10-24 · 15 minReconstruct an agent-assisted incident from evidenceA practical method for recovery, timelines, source and artifact identity, minimal reproductions and useful follow-up work.Read story ↗
GrantTap field guide · 2026-10-23 · 16 minMake mobile agent control readable and recoverableA practical accessibility workshop for meaningful statuses, stable scrolling, large text, voice drafts, notifications and deliberate actions.Read story ↗
GrantTap field guide · 2026-10-22 · 15 minUpgrade a dependency with an agent and keep the project reproducibleFrom the reason for an update through lockfile review, clean installation, application checks and a verified rollback baseline.Read story ↗
GrantTap field guide · 2026-10-21 · 15 minDesign MCP tools that remain understandable after a failed callA tool-contract workshop covering explicit targets, structured results, failures, repeated requests, cancellation and authorization.Read story ↗
GrantTap field guide · 2026-10-20 · 15 minUse Graph and Health to investigate a code boundaryA practical investigation method: report identity, structural hypotheses, source inspection and a controlled integration check.Read story ↗
Practical engineering with agents · 2026-10-19 · 15 minAPI compatibility when clients and servers update separatelyTest the supported version pairs, unknown states, queued data, capabilities and rollout order instead of trusting the newest local pair.Read story ↗
Practical engineering with agents · 2026-10-18 · 15 minDatabase migrations with an agent: rehearse the transition and recoveryPreserved values, concurrent writers, resumable backfills, integrity checks and a demonstrated restore before a live change.Read story ↗
Practical engineering with agents · 2026-10-17 · 15 minConfigure an agent project without putting secrets in the chatA reproducible environment contract: required fields, validation, precedence, process lifecycle and separate authentication layers.Read story ↗
Continuity · 2026-10-16 · 12 minMove a Task to another agent without losing workA Task can continue with another agent or on another computer. Uncommitted changes now raise a warning instead of blocking the move, a local checkpoint can keep them, and the destination confirms before the new chat opens.Read story ↗
GrantTap guide · 2026-10-16 · 10 minMessages that wait for the computerA queued request can wait on the encrypted relay until its computer reconnects, even with the phone closed. What each mark on your message means, what the relay can see, and where waiting stops.Read story ↗
Practical engineering with agents · 2026-10-15 · 15 minLong builds without guesswork: processes, pipelines and artifactsKeep the original run identifiable, catch intermediate failures, recover after interruption and verify the actual artifact.Read story ↗
Practical engineering with agents · 2026-10-14 · 15 minTime zones and calendar rules: give an agent the right temporal contractA practical workshop on publication dates, expiry, recurrence, fixed clocks and boundary tests across devices and locales.Read story ↗Practical agent workflows · 2026-10-13 · 15 minLight, Dark and Auto: a website theme that keeps its promiseGrantTap’s website as a practical case study: system changes, explicit overrides, storage failures, first paint, cross-tab updates and readable technical pages.Read story ↗
Practical agent workflows · 2026-10-12 · 15 minThe agent changed the file. Why is the app still old?Trace the real workspace, Git revision, local edits, build inputs and deployed artifact before repeating a correct change in the wrong checkout.Read story ↗
Practical agent workflows · 2026-10-11 · 16 minAgent permission denied on macOS: find the actual boundaryA focused file-access investigation: identify the path, process, owner, system permission and provider policy before changing access.Read story ↗
Engineering · 2026-10-10 · 15 minCortex Loom inside GrantTap: where token savings come fromA real Engine experiment reduced repeated source evidence by 56.2%, with zero omitted items. How citations, budgets and scoped delivery make the result useful.Read story ↗
Continuity · 2026-10-10 · 11 minProject Mesh: preserve work instead of doing it twiceObserved claims, scoped memory and a stable Task help agents coordinate and continue across models and computers. We verified failed-attempt persistence through an Engine restart.Read story ↗
Governance · 2026-10-10 · 11 minGrantTap governance: a decision that reaches the toolPolicy precedence, exact approvals and visible enforcement coverage. Real Engine and native hook checks show how the computer enforces the user's boundaries.Read story ↗
Practical agent workflows · 2026-10-09 · 15 minWhy a sleeping Mac does not answer your coding-agent messageDistinguish display sleep, system sleep, relay storage, computer receipt and agent activity before retrying the original message.Read story ↗
Practical agent workflows · 2026-10-08 · 15 minMCP, skills and hooks: choose the right agent extensionA working map of tools, reusable instructions and lifecycle events, with a small fixture that separates availability, permission and actual use.Read story ↗
GrantTap guide · 2026-10-05 · 7 minHow to control Claude Code and Codex from iPhone with GrantTapA practical path from local setup and device pairing to one Task view, approvals, and honest handoff across coding agents.Read story ↗
GrantTap guide · 2026-10-05 · 8 minGrantTap Mesh Governance: allow, ask, and deny for local coding agentsHow Project policy reaches each computer, what coverage means, and how to verify a rule before trusting it with a real Task.Read story ↗
AI security news · 2026-10-05 · 7 minMicrosoft Agent 365 expands MCP tool governance: what local coding agents should learnMicrosoft's September 2026 controls separate tool inventory, administrator approval, and runtime policy. Here is the narrower, practical GrantTap connection.Read story ↗
AI security news · 2026-10-05 · 7 minAWS AgentCore temporal policies make agent history part of authorizationAWS now describes stateful gateway checks for tool sequences, cumulative exposure, and human approval. We map the lesson to GrantTap's local boundary without claiming feature parity.Read story ↗
AI security news · 2026-10-05 · 7 minAnthropic's agent security disclosure: the boundary must exist outside the promptAnthropic described evaluation incidents and stronger containment in August 2026. The practical lesson for local coding agents is to verify host and network boundaries.Read story ↗
Architecture · 2026-10-05 · 7 minLocal, cloud, or hybrid: map an agent's real data boundaryThe location of the phone is the least useful architecture fact. Follow code, model context, tools, and control decisions instead.Read story ↗
Evidence · 2026-10-05 · 7 minWhat agent usage metrics can prove—and what they cannotToken counts, plan limits, tool calls, cost, and useful output are different measurements. Here is a more honest dashboard vocabulary.Read story ↗
Continuity · 2026-10-04 · 7 minA task can outlive an agent session. Here is what must travel with itSession resume, cloud handoff, and a cross-provider Task solve different continuity problems.Read story ↗
Governance · 2026-10-04 · 7 minAn agent approval is only as strong as its execution boundaryHow to tell a displayed decision from an enforced one across local, cloud, and hybrid coding agents.Read story ↗
Engineering · 2026-10-04 · 8 minCortex Loom: fewer context tokens, with the gaps visibleA task-aware evidence packet can be much smaller than a folder dump. The benchmark also shows why token savings alone are not the goal.Read story ↗
Comparison · 2026-10-03 · 7 minChoosing a mobile coding-agent workflow by the work it must finishA practical comparison of Claude Code, Codex, Cursor, and GrantTap across five real moments away from the desk.Read story ↗
Field notes · 2026-10-03 · 7 minWhat the AWS Summit Tel Aviv agenda says about agentsThe September 2026 agenda paired building and running agents with optimization, observability, and evaluation.Read story ↗
Governance · 2026-10-03 · 7 minAgent governance must reach the computerAn allow/ask/deny switch is only the beginning. Effective policy, host enforcement, and observed use need separate evidence.Read story ↗
Landscape · 2026-10-03 · 8 minCoding agents on your phone: what each approach controlsClaude Code, Codex, and Cursor already offer mobile paths. The useful comparison is where the agent runs and what you can verify.Read story ↗
Product thinking · 2026-10-03 · 7 minWhy GrantTap is a control center for local agentsThe task stays visible while providers, sessions, and computers change. Here is the product decision behind that design.Read story ↗
Getting started · 2026-09-24 · 7 minConnect your iPhone without another accountThe computer shows the pairing QR. The phone scans it. A Mesh invitation is a different action.Read story ↗
Workflow · 2026-09-24 · 6 minOne Task, several executionsHow a stable Task keeps its history when work moves between a provider, a computer, or a native session.Read story ↗
Mesh · 2026-09-24 · 6 minLinked Mesh spaces keep separate permissionsOne Mesh can link several repositories; links between Mesh spaces do not merge access.Read story ↗
Architecture · 2026-09-24 · 7 minA graph should answer a code questionHow to read Weavatrix evidence, full-screen code towers, and the difference between a report and a repository binding.Read story ↗
Governance · 2026-09-24 · 7 minConfigured, available, used: three different factsA practical way to read MCP and skill status without confusing a catalog entry with working execution.Read story ↗