# Five things featkpr does today, each shown on BookStack As of 27 Sep 2026. Connecting your own code is on the roadmap; until then, a demo call shows all five. - **Permission tests** (live): Tests per route and per role, each sent again as a user without permission. It must be refused. Proof: 269 of 304 BookStack tests refused to the wrong user. https://featkpr.com/services/permission-tests.md - **Feature map** (live): Modules and features named from the repository, each with the routes and screens behind it. Proof: 307 features in 12 modules named on BookStack. https://featkpr.com/services/feature-map.md - **User journeys** (live): Each way a person reaches a goal, screen by screen, for you to keep or drop. Proof: 163 paths drafted to 69 goals on BookStack. https://featkpr.com/services/user-journeys.md - **Generated tests** (live): A test for each route and role on the map, none written by hand, with what it sent and what came back. Proof: 481 tests written so far; 304 ran and passed in the last run on BookStack. https://featkpr.com/services/playwright-tests.md - **Change impact** (building): For a pull request: the routes, features and tests it touches, so only those re-run. Shown on a recorded run so far. Proof: shown on BookStack pull request #6108, in a recorded run. https://featkpr.com/services/change-impact.md ## Everything, by state ### Live (10) - **Feature map from the code**: Modules and features named from the repository, per product. Receipt: 307 features in 12 modules on BookStack · 20 on Shlink's frozen map. When: since 22 Sep. Source: ST · CTX. - **Crawl with real screens**: Every page per role, forms filled and sent, on a throwaway copy. Receipt: 167 screens · 8 forms sent · admin, viewer and anonymous. When: since 25 Sep. Source: ST 25 Sep. - **Paths to goals**: Each journey with its screens and a drafted goal, for you to keep or drop. Receipt: 163 paths to 69 goals · 28 goals drafted for $0.04. When: since 25 Sep. Source: ST 25 Sep. - **Scenario trees**: Who, with which setting and input, gets which outcome, per feature. Receipt: a test written for every leaf: 481 on BookStack. When: since 22 Sep. Source: DS. - **Generated tests, with the wrong-user check**: Tests written from the map that run, each re-sent as the wrong user. Receipt: 304 of 304 pass · 269 refused to the wrong user. When: since 26 Sep. Source: ST 26 Sep. - **Runs, with evidence**: Every test: what was sent, what came back, why it failed, a picture. Receipt: 256 pictures in the latest run. When: since 26 Sep. Source: ST 26 Sep. - **From-zero audit**: Every stage checked against a truth it never reads, each miss named. Receipt: BookStack 605 of 2,372 missed · Shlink 87 of 169. When: since 26 Sep · run by us today. Source: HA. - **Merge-request impact**: What one change reaches, before anyone runs anything. Receipt: shown on BookStack pull request #6108, in a recorded run. When: a recorded run. Source: ST 26 Sep. - **Several products, one place**: A product switcher and a sign-in in front of it. Receipt: BookStack and Shlink. When: since 24 Sep. Source: ST 24 Sep. - **Pages for agents**: Every page of this site has a markdown twin, plus llms.txt. Receipt: send Accept: text/markdown to any page. When: live. Source: the /agents page of this site. ### Building (3) - **Goals keyed by their outcome**: A goal described in 12 ways: its structure, the input it needs, third parties, computations. Receipt: the prototype reads all 343 routes in 0.43 s. When: week 0, from 28 Sep. Source: DS · TL. - **A crawl that edits and deletes what it made**: Covers the edit and delete forms. Every change versioned and explained, released as a draft after your approval. Receipt: today 4 of 87 forms are sent. When: week 0. Source: TL · HA · DEC. - **Drafts checked by several models**: One model drafts each goal, and cheaper models verify it. Receipt: 96% pass · $0.17 per 100 goals · experiment of 27 Sep cost $1.10. When: verify experiment running. Source: ME · DEC. ### Planned (13) - **Complexity, drafted by featkpr**: Paths, inputs, preconditions, external calls, waiting and roles. featkpr drafts, you correct. Receipt: no owner ranking: decided 27 Sep. When: week 1, from 5 Oct. Source: TL · DEC. - **Effects in every run**: Mail sent, calls leaving the host, downloads, rows changed. Receipt: today 0 of 33 effects are recorded. When: week 1. Source: TL · HA. - **Tracing with no code change**: OpenTelemetry on the running product, with nothing added to its code. Receipt: decided 27 Sep. When: week 1. Source: TL · DEC. - **Release flow: where it is proven**: Trunk, GitHub flow, GitFlow, stable and unstable, environments. A Kargo adapter reads your stages. Receipt: decided 27 Sep. When: week 2, from 12 Oct. Source: TL · DEC. - **Stays up when the forge doesn't**: A hardened queue, a mirror of the repository and an operator screen. Receipt: today an outage fails an item after about 3 h 15 min. When: week 2. Source: DS · TL. - **Goals view**: How complex each goal is and what shape it makes: straight line or branching. Receipt: design pass first. When: weeks 3–4, from 19 Oct. Source: TL. - **Whole-system tracing**: CPU, memory, queries, cache, queues and logs during every state change. Receipt: decided 27 Sep. When: weeks 3–4. Source: TL · DEC. - **Your product's own telemetry**: Whatever it already emits, read on any stack. Receipt: decided 27 Sep. When: weeks 3–4. Source: TL · DEC. - **Merge-request verdict**: A verdict on a change, from the goals it proved and broke. Receipt: impact is live; the verdict is not. When: weeks 3–4. Source: TL. - **Branching paths and role views**: Ways in, choke points and one view per role. Receipt: 144 routes × 8 roles fold into 22 patterns. When: after the goal ledger. Source: BR2. - **Questions as a floating dot**: Quick decisions in a corner, each showing what it settles. Receipt: raised 26 Sep. When: next, not dated. Source: DS. - **Analytics and performance**: A feature-flag adapter (GrowthBook, OpenFeature), web vitals, Lighthouse. Receipt: not on the timeline yet. When: next, not dated. Source: DS. - **A read-only server for agents**: The map, queried by your coding agent. Receipt: not shipped. When: not dated. Source: the /agents page of this site. ### Ideas (5) - **A demo environment per feature**: A working copy of one feature or feature set, with a read-only link for a client or a lead. Receipt: half runs today: a throwaway copy per commit, one user per role. When: raised by Tim, 27 Sep. Source: DEC · ST. - **A guide to your product, from the map**: Every goal as steps with screens, who can do it, where it fails. Receipt: the in-app tour is live; this is not. When: raised by Tim, 21 Sep. Source: CTX. - **A product wiki and a flags hub**: What the product does and what is switched on, in one place. Receipt: coming-soon surfaces. When: raised by Tim, 22 Sep. Source: CTX. - **Findings filed with maintainers**: Bugs the map finds in open-source products, filed and published with the reply. Receipt: nothing filed yet. When: raised by Tim, 22 Sep. Source: the /findings page of this site. - **Flows at every scale**: A conversation beside a canvas that draws the flow you are talking about. Receipt: raised 25 Sep. When: raised by Tim, 25 Sep. Source: ST 25 Sep. Source keys: TL = our plan of 27 Sep 2026; DS = design notes of 26 Sep 2026; BR2 = design notes of 27 Sep 2026; ME = the model experiment of 27 Sep 2026; HA = the from-zero audit of 26 Sep 2026; DEC = decisions of 27 Sep 2026; ST = our build log; CTX = our project notes; PK = our package plan. --- The page this twin stands for: https://featkpr.com/services. Every page on this site has a `.md` twin, and answers `Accept: text/markdown`.