Signed-off-by: Nikos Mavrogiannopoulos <n.mavrogiannopoulos@gmail.com>
20 KiB
AGENTS.md — ocserv AI Agent Guide
This file is the single source of guidance for AI coding agents (Claude Code, Codex, Copilot, Cursor, and others) working in this repository. Read it fully before writing or modifying any code.
Security Disclosure — Read This First
If you are helping investigate a potential security vulnerability:
- Stop. Do not open a public issue or merge request.
- Direct the contributor to create a confidential issue here: https://gitlab.com/openconnect/ocserv/-/issues/new?type=ISSUE&initialCreationContext=list-route On that page, check "This issue is confidential" before submitting.
- Do not draft a public patch until the maintainers have confirmed the issue and coordinated a fix.
The bar for using this path is suspicion, not certainty. If you are unsure whether something is a vulnerability, ask the human contributor before posting anything publicly. Unverified public reports consume significant maintainer time. Only report if you can describe a concrete impact; do not report theoretical weaknesses without a demonstrated exploit path.
Architecture — The Invariant You Must Not Violate
ocserv enforces security through process isolation. There are three processes:
| Process | Privilege | Responsibilities |
|---|---|---|
main (main.c, main-*.c) |
root | TCP/UDP listeners, TUN devices, process lifecycle, IP allocation |
sec-mod (sec-mod.c, sec-mod-*.c) |
root | Authentication, private keys, session state, PAM, accounting |
worker (worker.c, worker-*.c) |
unprivileged + seccomp | TLS/DTLS per client, VPN traffic bridging |
Workers communicate with main and sec-mod exclusively through Unix sockets using
protobuf IPC (defined in src/ipc.proto and src/ctl.proto). Workers have no
direct access to credentials, keys, or privileged system calls.
Hard rule: No patch may move credential handling into a worker, grant a worker direct filesystem or socket access outside its seccomp profile, or silently collapse any of these boundaries. If a proposed change requires crossing a privilege boundary, flag it explicitly in the MR for maintainer review — do not proceed without acknowledgment.
Project Overview
OpenConnect VPN Server (ocserv) is an SSL VPN server compatible with the OpenConnect protocol and Cisco AnyConnect. It is a security-critical network daemon. Linux is the intended deployment platform. BSD compatibility (FreeBSD, OpenBSD) is maintained on a best-effort basis: we try not to break it, but features will not be rejected solely because they do not work on BSD.
Requirements-First Workflow
doc/requirements/ is the normative description of what ocserv must do (see
doc/requirements/README.md for the document map, ID scheme, and per-requirement
format). For any change that alters observable behavior — new feature, bug fix, or
behavior-changing refactor — work in this order:
- Requirement first.
- Find the requirement(s) covering the area you're changing (search
doc/requirements/for the relevantREQ-*IDs, or use the document map to find the right file by process/subsystem). - If your change alters what an existing requirement describes, update that
requirement first, re-applying the protocol in
contrib/ai/protocols/that generated its document so the new entry matches the rest of the file. Do not leave a requirement describing the old behavior once the code changes — but do not break other requirements or use-cases in the process (see below). If a requirement is independently wrong and needs revising for reasons unrelated to this change, do that in its own dedicated merge request instead. - If no requirement covers the new behavior, add one in the appropriate document, following its existing ID prefix, category tags, and per-requirement format.
- For bug fixes: if the bug is a violation of an existing requirement, cite its ID in the commit/MR. If the bug reveals a gap in the requirements, extend or add a requirement describing the correct behavior before fixing the code.
- Find the requirement(s) covering the area you're changing (search
- Tests second. Write or update the positive and negative tests implied by the requirement's Acceptance criteria (see "Testing New Functionality" below) before touching implementation code. For bug fixes, confirm the new test reproduces the bug and fails against the unmodified code.
- Code last. Implement the change so the new/updated tests pass, and so the
code, the requirement, and
doc/ocserv.8.md/doc/sample.config(where applicable) all agree.
Do not jump straight to step 3 — a code change with no corresponding requirement update is incomplete, even if it builds and passes existing tests.
Do not break existing requirements or use-cases
Before submitting a merge request, check that the change does not silently invalidate requirements or use-cases other than the one you set out to change:
- Search
doc/requirements/forREQ-*/AC-*/OC-*entries citing the files, functions, or config options you touched, and confirm each still holds. If one no longer holds, either your change is wrong, or that requirement was wrong to begin with (mark itREVIEWif unsure which) — never leave code and aDERIVEDrequirement contradicting each other. - A requirement that's wrong for reasons unrelated to your change belongs in its own dedicated merge request — with rationale and evidence it doesn't break other use-cases — not bundled into this one.
- Run the full local test suite (
ninja -C build test), not just the test for the area you changed — requirements frequently span files (e.g. config scope/ inheritance, IPC message contracts) and a regression often shows up as a failure in a test you didn't expect to touch. - If your change affects config scope, inheritance, or vhost behavior, also run
tests/check-config-scope.py(seedoc/requirements/internal/config.md).
Build System
The project uses meson.
# Initial setup
git submodule update --init
meson setup build
cd build && ninja
# Run all tests
ninja -C build test
# Run a specific test
meson test -C build test-pass
Common Build Options
| Option | Default | Description |
|---|---|---|
-Doidc-auth=enabled |
disabled | OpenID Connect authentication |
-Dpam=disabled |
auto | PAM authentication |
-Dradius=disabled |
auto | RADIUS authentication/accounting |
-Dseccomp=disabled |
auto | seccomp worker isolation |
-Dwith-werror=true |
false | Treat warnings as errors |
-Db_coverage=true |
false | Enable gcov coverage |
# Minimal build for fast iteration
meson setup build -Dradius=disabled -Dpam=disabled \
-Dseccomp=disabled -Danyconnect-compat=disabled
# View or change options
meson configure build
Build output: build/. Test logs: build/meson-logs/testlog.txt.
Build Dependencies
Debian/Ubuntu:
apt-get install -y build-essential meson ninja-build pkg-config \
libgnutls28-dev libev-dev libreadline-dev libtasn1-bin
# Optional:
apt-get install -y libpam0g-dev libseccomp-dev libradcli-dev \
libcurl4-gnutls-dev libprotobuf-c-dev libtalloc-dev libllhttp-dev \
protobuf-c-compiler gperf libsocket-wrapper libpam-wrapper \
libnss-wrapper libuid-wrapper freeradius
Fedora/RHEL:
yum install -y meson ninja-build gcc pkgconf-pkg-config \
gnutls-devel libev-devel readline-devel libtasn1-tools
# Optional:
yum install -y pam-devel libseccomp-devel radcli-devel libcurl-devel \
protobuf-c-devel libtalloc-devel llhttp-devel protobuf-c gperf \
socket_wrapper pam_wrapper nss_wrapper uid_wrapper freeradius
Testing
# All tests
ninja -C build test
# Verbose output
ninja -C build test -- --verbose
Test Environment Variables
VERBOSE=1 # Enable verbose server output
COVERAGE=1 # Disable worker isolation for coverage
ISOLATE_WORKERS=false # Disable seccomp (needed for ASAN/coverage)
DISABLE_ASAN_BROKEN_TESTS=1 # Skip ldpreload-based tests (incompatible with ASAN)
Test Structure
tests/— Shell test scripts and supporting filestests/data/— Config templates with placeholders (@USERNAME@,@PORT@,@SRCDIR@) substituted byupdate_config()intests/common.shtests/certs/— Test certificates and keystests/common.sh— Shared utilities (requires theopenconnectclient)tests/*.c— Unit tests for specific components
New functionality requires both positive and negative test cases.
Register new tests in tests/meson.build.
Code Organization
Source Layout (src/)
- Main server:
main.c,main-*.c - Security module:
sec-mod.c,sec-mod-*.c - Worker:
worker.c,worker-*.c - Auth modules:
auth/— PAM, GSSAPI, RADIUS, OIDC, plain password - Accounting modules:
acct/— PAM, RADIUS - Configuration:
config.c(global options),subconfig.c(per-module bracketed options) - TLS:
tlslib.c— GnuTLS wrapper - TUN device:
tun.c - IP management:
ip-lease.c,ip-util.c - IPC definitions:
ipc.proto,ctl.proto(protobuf) - Control tool:
occtl/ - Password tool:
ocpasswd/
Key Headers
vpn.h— Main VPN data structuresmain.h— Main server structures (main_server_st,proc_st)sec-mod.h— Security module structuresworker.h— Worker process structurescommon-config.h— Config structs (cfg_st,perm_cfg_st, per-module structs)defs.h— IPC command enum (cmd_request_t), error codes, log levels, auth statesvhost.h— Virtual host config; use macrosGETVHOST(s),GETCONFIG(s),GETPCONFIG(s)log.h— Logging:mslog()in main,oclog()in worker,seclog()in sec-mod
External Libraries (src/)
ccan/— Hash tables, lists, talloc, and more (check here before adding utilities)inih/— INI file parserllhttp/— HTTP parserpcl/— Portable Coroutine Libraryprotobuf/— Protocol Buffers implementation in Cgnulib/— Provides thecloexecmodule only; do not import additional gnulib modules
User Documentation (doc/)
ocserv.8.md— Main server man page; authoritative description of all configuration options, their defaults, and expected behaviorsample.config— Annotated reference configuration; authoritative for option scope and default valuesocpasswd.8.md— Password management tool man pageocctl.8.md— Control and monitoring tool man page
When investigating a bug or implementing a feature, consult these files to establish what behavior is documented to the administrator. The code must match the documentation, or the documentation must be updated alongside the code change.
Architecture & IPC Communication
For full diagrams and protocol descriptions, read doc/design.md — it is the
authoritative reference for what data crosses which process boundary.
Authentication Sequence
worker -> sec-mod: SEC_AUTH_INIT (username / password / cert)
sec-mod -> worker: SEC_AUTH_REP (new SID assigned)
worker -> sec-mod: SEC_AUTH_CONT (SID + password for multi-factor)
sec-mod -> worker: SEC_AUTH_REP (OK or FAILED)
worker -> main: AUTH_COOKIE_REQ (SID)
main -> sec-mod: SECM_SESSION_OPEN (SID)
sec-mod -> main: SECM_SESSION_REPLY (user config)
main -> worker: AUTH_COOKIE_REP (TUN device, config)
worker -> main: SESSION_INFO (periodic updates)
Key Concepts
- SID: Assigned by sec-mod on
SEC_AUTH_INIT. Thewebvpncontextcookie sent to clients is only for resuming fully authenticated sessions — new connections must always start withSEC_AUTH_INIT. - Cookie: Auth token valid for
cookie-timeout; enables mobile roaming. - TUN device: Forwarded from main to worker after successful authentication.
- DTLS: UDP channel in parallel to TLS for better throughput.
Modifying IPC
If your change touches src/ipc.proto or src/ctl.proto, regenerate the C bindings:
protoc-c --c_out=src/ src/ipc.proto
protoc-c --c_out=src/ src/ctl.proto
Do not edit the generated *.pb-c.c / *.pb-c.h files by hand.
Development Guidelines
Canonical Technology Choices
One tool for each job. These decisions are requirements, not preferences — the
normative form with acceptance criteria is in
doc/requirements/internal/general.md. Departing from any of them requires a
design-discussion issue and explicit maintainer approval.
| Concern | Tool | Requirement |
|---|---|---|
| Memory management | talloc | Use talloc_zero, talloc_strdup, talloc_array, etc. everywhere. Exception: gnutls_malloc()/gnutls_free() only for GnuTLS-owned memory — never mix the two for the same allocation. Error paths: goto cleanup with a single freeing label. See REQ-GEN-TECH-001. |
| Cryptography | GnuTLS / nettle | No OpenSSL. Call GnuTLS through src/tlslib.c wrappers — not directly from worker or main code. See REQ-GEN-TECH-002. |
| IPC serialization | protobuf-c | All cross-process messages are defined in src/ipc.proto and src/ctl.proto. Regenerate C bindings with protoc-c after any .proto edit; never hand-edit generated files. See REQ-GEN-TECH-003. |
| Configuration parsing | inih (INI) | Flat key = value pairs; bracketed sub-sections for per-module options (auth = pam[...]). No new structured sub-formats (no JSON, YAML, or additional config files for things expressible as key=value). See REQ-GEN-TECH-004. |
| New external dependency | requires approval | Check src/ccan/ first; consider a trivial inline implementation; if neither suffices, open a design-discussion issue before adding. See REQ-GEN-TECH-005. |
Design Principles
Locality of complexity. A feature belongs in as few files as possible. The ideal: a reviewer understands it by reading one file, not tracing a chain across ten. New cross-module state, callbacks, and "utility" files are a design smell — if your change requires them, reconsider the module boundary first. When complexity must be hidden, hide it inside a module with a clean header interface; do not scatter it.
Resist dependency growth. See REQ-GEN-TECH-005 for the normative rule.
In short: check src/ccan/ first, then consider a trivial inline implementation;
only then open a design-discussion issue for a new external library.
Configuration stays in INI. See REQ-GEN-TECH-004 for the normative rule.
Explicit over implicit. No auto-discovery, no runtime plugin loading, no silent
defaults that change behavior. Every configurable behavior must be expressible in
doc/sample.config.
Code Style
See REQ-GEN-STYLE-001 and REQ-GEN-STYLE-002 for the normative rules. In brief:
- C99 standard; Linux kernel coding style (tabs, 8-space tab width, 80-column limit)
clang-format --dry-run -Werrormust pass on all files undersrc/andtests/- Header guards:
#ifndef FILENAME_H/#define FILENAME_H/#endif /* FILENAME_H */ #ifdefin function bodies: ≤ 1 nesting level, ≤ 5 lines per branch; optional features use the stub pattern;#endifalways annotated- Comments: Prefer self-documenting code. Add a comment only when the why is non-obvious: a hidden constraint, a protocol expectation, or a workaround. Do not comment what the code does; well-named identifiers already do that.
Testing New Functionality
See REQ-GEN-TEST-001 for the normative rule. In brief:
- Write a positive test and a negative test for every feature or fix.
- For security-relevant code (auth, cookies, IPC validation) write the negative test first.
- For bug fixes, confirm the reproducing test fails before applying the fix.
- Tests must be self-diagnosing — failure explainable from output alone; shell tests
print
"FAIL: expected X, got Y"; C unit tests print the failing condition and values. - Register all new tests in
tests/meson.build.
Adding Configuration Options
- Global options:
src/config.c - Per auth/acct module options: struct in
src/common-config.h+ parser insrc/subconfig.c
Implementing a New Auth Module
Follow the auth_mod_st vtable in src/sec-mod-auth.h. Register in src/sec-mod.c.
See src/auth/plain.c or src/auth/pam.c for reference.
Git Commits
- Every commit must have a
Signed-off-by: Name <email>line (DCO requirement) - Use
Resolves: #NNNon the fixing commit;Relates: #NNNon related commits (e.g., tests)
Platform Portability
See REQ-GEN-COMPAT-001 for the normative rule. In brief: Linux is the primary
target; BSD is best-effort. Guard Linux-specific code with #ifdef __linux__. Note
in doc/ocserv.8.md any option whose behavior differs on BSD (e.g. no live reload
without procfs).
Module-specific Documentation
Before working on a subsystem, read the relevant doc:
| If your change touches... | Read before coding |
|---|---|
| IPC / process communication | doc/design.md |
| RADIUS auth or accounting | doc/README-radius.md |
| OpenID Connect auth | doc/README-oidc.md |
| Cisco SVC / AnyConnect compatibility | doc/README-cisco-svc.md |
Contribution Checklist
Workflow note: If you are helping a contributor report a bug or submit a fix, direct them to open a GitLab merge request rather than attach a patch to the issue. See
CONTRIBUTING.mdfor the expected workflow.
Agent-runnable — verify before declaring a change complete
- Relevant
doc/requirements/entries added or updated before the code change (see "Requirements-First Workflow" above) clang-format --dry-run -Werrorpasses on every modified file undersrc/andtests/ninja -C buildsucceeds- Relevant test passes:
meson test -C build <test-name> - Full suite passes:
ninja -C build test(no regressions in unrelated tests) - Existing
doc/requirements/entries touching the changed files/functions still hold (none silently contradicted) - Every commit has
Signed-off-by:
Human-judgment required — flag in the MR, do not decide unilaterally
- Any change that crosses a process privilege boundary
- New syscall in the worker path (requires seccomp filter review by maintainer)
- New auth method — design must be discussed before implementation
- TLS or DTLS behavior changes (cipher selection, version negotiation, certificate handling)
- Changes to cookie or SID handling
CI/CD
CI is configured in .gitlab-ci.yml. Tests run on Debian, Ubuntu, Fedora, CentOS, Alpine,
and includes a cross-compilation job (i386/Debian). CI also checks:
Signed-off-byon all commitsclang-formaton allsrc/andtests/files (excluding generated and vendored code)
To run CI jobs locally with podman/docker:
./run-ci-local.sh quick # Fast validation
./run-ci-local.sh debian # Single platform
./run-ci-local.sh core # Core platforms
./run-ci-local.sh all # Full matrix (~1–2 hours)
Personas
For extended AI-assisted workflows, load the appropriate persona as a system prompt prefix before starting work.
-
Maintainers (bug investigation, code review, refactoring, design):
contrib/ai/personas/ocserv-core-dev.md -
External contributors (feature additions, bug fixes, security fixes):
contrib/ai/personas/ocserv-contributor.md -
Security audits (vulnerability discovery, threat modeling, secure design review):
contrib/ai/personas/ocserv-security-auditor.md
These personas embed project-specific protocols for anti-hallucination, memory safety, security vulnerability taxonomy, exhaustive path tracing, stack lifetime hazards, and self-verification.